Software development has rapidly evolved from explicit, instruction-based programming to integrating the non-deterministic reasoning capabilities of Large Language Models.
Now, we face the next massive paradigm shift: moving from passive AI tools to Agentic AI.
In this workshop, you will learn everything you need to build these autonomous systems using Java, Spring AI, the Model Context Protocol (MCP) and more. We will move beyond basic prompting to master chat memory, RAG, tool integration, and agentic patterns. Most importantly, we address the "day two" challenges of production: ensuring your agents are observable, secure, and reliable. By the end of the session, you will have built a multi-agent system capable of solving complex, high-level tasks.
Join us and transform your applications from simple interfaces into proactive digital colleagues. ","fieldGroup":null,"fieldGroupValues":[],"id":24229,"keywords":[{"name":"agentic"},{"name":"java"},{"name":"autonomous"},{"name":"spring"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#00FF00","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":968,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":2,"name":"Hands-on Lab","pause":false,"scheduleInSmallRoom":true,"slug":"lab18"},"speakers":[{"anonymizedBio":null,"bio":"
Timo Salm is a Principal Solutions Engineer at VMware Tanzu by Broadcom with over a decade of experience in customer-facing roles, modern applications, DevSecOps, and AI. In his roles, he ensures that the most strategic customers in the EMEA region achieve their goals with VMware Tanzu's Spring, developer platform, AI, and data products. Another essential part of his role is continuously learning through experimenting hands-on with innovations and sharing the outcomes with colleagues, customers, and the community via, for example, conferences. Before Timo joined Pivotal, which VMware and now Broadcom acquired, he worked for consulting firms in the automotive industry as a software architect and full-stack developer.","blueskyUsername":null,"company":"Broadcom","countryName":"Germany","firstName":"Timo","fullName":"Timo Salm","id":7606,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-93dfac6b-7673-4fca-81c6-aa9fb4a0c084.jpg","lastName":"Salm","linkedInUsername":"timosalm","mastodonUsername":null,"twitterHandle":"salmto"},{"anonymizedBio":null,"bio":"Sandra Ahlgrimm is a Senior Developer Advocate for GitHub at Microsoft, specializing in supporting Java Developers. With over a decade of experience as a Java developer, she brings a wealth of knowledge to her role. Sandra is passionate about containers and has recently learned to love AI.
As a leader in the tech community, Sandra actively contributes to the Berlin Java User Group (JUG) and the Berlin Docker MeetUp. Her expertise extends beyond coding; she focuses on LangChain4j integrations and serves as the primary point of contact for developer feedback related to Java in Visual Studio Code (VS Code) and GitHub Copilot in IntelliJ. Additionally, her interest in performance and event-driven architectures led to her involvement with native images. Therefore, Sandra represents Microsoft on the GraalVM Program Advisory Board.","blueskyUsername":"","company":"Microsoft / GitHub","countryName":"Germany","firstName":"Sandra","fullName":"Sandra Ahlgrimm","id":7617,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-055a7a76-5e5c-4ed2-a7af-f624fb91c129.jpg","lastName":"Ahlgrimm","linkedInUsername":"sandraahlgrimm","mastodonUsername":"","twitterHandle":"skriemhild"}],"summary":"This workshop teaches building autonomous agentic AI systems with Java, Spring AI, and MCP. It covers prompting, chat memory, RAG, tool integration, and agentic patterns, plus production concerns like observability, security, and reliability. Attendees will create a multi-agent system for complex tasks.","timeSlots":[],"title":"Building Autonomous AI Agents with Spring","totalFavourites":12,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Most developers know OpenAPI as the standard for describing and generating code around synchronous APIs like REST, a single spec that produces client code, server stubs, and (Swagger) docs, keeping everything originating from the same spec.
AsyncAPI brings that spec-first approach to event-driven systems: message brokers, topics, consumers and producers, described in a machine-readable definition that tooling can act on.
In this tools-in-action session, we'll use AsyncAPI to generate a working Spring Boot producer and consumer for Kafka, including the message DTOs and the API documentation straight from the spec. No hand-written boilerplate. We define the spec, generate both sides of the conversation, and watch a message flow end-to-end between them.
After this session, you'll have a clear understanding of what AsyncAPI is and how it relates to OpenAPI, how spec-first generation removes hand-written boilerplate and drift between producer, consumer, and docs, and enough hands-on familiarity with the tooling: spec structure, generator templates, to try it on your own event-driven services.","fieldGroup":null,"fieldGroupValues":[],"id":24234,"keywords":[{"name":"kafka"},{"name":"asyncapi"},{"name":"event-driven"},{"name":"openapi"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Michel Schudel is a passionate software engineer. He has over 20 years of hands-on experience in IT development, worked with a variety of languages, tools and infrastructures, ranging from Turbo Pascal monoliths running in a basement to JVM-based microservices in the cloud. During that time, he has seen the good, the bad and the ugly in development land. As a regular speaker at conferences like Devoxx and J-Fall, he takes every opportunity to share the things he learned, whether it be technical and non-technical subjects, with his peers.
","blueskyUsername":null,"company":"Craftsmen","countryName":"Netherlands","firstName":"Michel","fullName":"Michel Schudel","id":24169,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GgNY5OcMk2ghQnNGYDAQBxLIE4ofeltIbvzAK_s=s96-c","lastName":"Schudel","linkedInUsername":"MichelSchudel","mastodonUsername":null,"twitterHandle":"MichelSchudel"}],"summary":"This session introduces AsyncAPI, the spec-first counterpart to OpenAPI for event-driven systems. Through a live demo, it generates a Spring Boot Kafka producer, consumer, DTOs, and docs from one AsyncAPI spec, showing how contract-first tooling eliminates boilerplate and keeps services and documentation aligned.","timeSlots":[],"title":"AsyncAPI in action: OpenAPI for event-driven applications","totalFavourites":16,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"JSON is ubiquitous in modern software, and many use cases are just too small to justify adding external dependencies.
To address these everyday use cases, a new JEP proposes a simple, incubating API for parsing, navigating, and generating JSON documents-both for Java developers and for the JDK itself.
While the Java ecosystem offers many libraries with JSON-processing capabilities, this built-in API is not intended to replace them, but rather to offer a low-friction option when an external library would be unnecessary or impractical.
In this talk, we will show how Data-Oriented Programming helps simplify the API, discuss its intended use cases, outline the design trade-offs, and demonstrate its capabilities.","fieldGroup":null,"fieldGroupValues":[],"id":23539,"keywords":[{"name":"java"},{"name":"data-oriented programming"},{"name":"api"},{"name":"json"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Viktor is a Software Architect in the Java Platform Group at Oracle.
For more than two decades, he has focused on enabling developers to solve problems with maximal productivity and maintainability.
Specializing in developer ergonomics in the realm of concurrent, parallel, and distributed programming; he has contributed to the standard libraries of several major programming languages, been involved in more than a dozen Open Source projects, and has spoken at numerous conferences and universities.","blueskyUsername":"","company":"Oracle","countryName":"Sweden","firstName":"Viktor","fullName":"Viktor Klang","id":23049,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTtdkY-Lks0gsRcvl05pqpW8-GOpvuVndojO99ZxUsSy0rgo=s96-c","lastName":"Klang","linkedInUsername":"","mastodonUsername":"","twitterHandle":""}],"summary":"A new JEP proposes a simple, incubating built-in Java API for parsing, navigating, and generating JSON for lightweight use cases where external libraries are unnecessary. The talk explains how Data-Oriented Programming shapes the API, its use cases, trade-offs, and capabilities.","timeSlots":[],"title":"Everyday JSON","totalFavourites":14,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Up until a couple of years ago, I was riding the waves of the latest technologies, shipping code, and mocking the "suits" who spent their days writing documentation that was never read. But then, the unthinkable happened. I joined a new organization and transitioned to the dark side. I became one of them: an Enterprise Architect. Or as I like to call myself: a fake enterprise architect.
I quickly found out that my new reality meant dealing with consultants pitching one-shot solutions as the ultimate silver bullet (pun intended), and uncovering the psychology of IT by learning why technical decisions are rarely about technology. At the same time, I wanted to stay in touch with the real world where frameworks evolve and pipelines run. But how on earth do you find the time for that?
This talk is a transparent look inside my shiny new office, commonly known as the Ivory Tower. Whether you're a developer trying to challenge corporate IT decisions, an architect struggling to explain them, or you just want to make fun of enterprise architects (the real ones, that is): this session is for a broad audience, sharing lessons learned trying to bridge the gap between two worlds that should never be disconnected.
Expect to walk away with:
`.","timeSlots":[],"title":"Why it takes 3000 commits to bring Valhalla's value objects to the JDK","totalFavourites":12,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"Historically, high-performance AI workloads have often depended on Python ecosystems, JNI-based bindings, or complex native integrations, leaving Java as a secondary choice. But the Java platform has changed dramatically. With the Foreign Function & Memory (FFM) API, the Vector API, and modern JVM capabilities, Java is becoming a first-class platform for AI development.
In this session, we'll show how the Foreign Function & Memory API and the Vector API enable a fully Java-native AI stack capable of running efficiently on both GPUs and CPUs. Using the latest release of Deep Netts, we'll demonstrate direct CUDA integration through the FFM API—without JNI or Python—and CPU acceleration using the Vector API for SIMD execution.
We'll present the architecture behind this approach, explain key implementation decisions, and compare it with traditional JNI- and Python-based solutions. The session also includes a real-world enterprise use case: AI-powered fraud detection, where low-latency inference and predictable performance are critical for banking and payment systems.
Whether you're interested in modern Java, high-performance computing, or enterprise AI, you'll leave with a practical understanding of how FFM, the Vector API, and GPU acceleration are transforming Java into a first-class platform for building and deploying production AI applications.","fieldGroup":null,"fieldGroupValues":[],"id":22947,"keywords":[{"name":"java"},{"name":"vector"},{"name":"ffm"},{"name":"gpu"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Zoran Sevarac is a seasoned software engineer, university professor of Artificial Intelligence, and passionate Java developer with more than two decades of experience in software engineering, AI research, and open-source innovation.
He is the creator of two pioneering AI development platforms built entirely in Java — Neuroph and Deep Netts — which made AI development more accessible to the Java community. His work bridges modern AI technologies with the reliability, performance, and scalability of the Java ecosystem.
A Duke’s Choice Award winner, Java Champion, and member of the JCP Executive Committee, Zoran has been instrumental in driving the evolution of AI support in Java. His mission is to simplify AI adoption, enable GPU acceleration through Project Panama and the Vector API, and bring high-performance, energy-efficient, and FinOps-friendly AI to enterprise systems.","blueskyUsername":"","company":"Deep Netts ","countryName":"Serbia","firstName":"Zoran","fullName":"Zoran Sevarac","id":23009,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-c08270b6-3020-4fc8-a6aa-395616d4bd0f.jpg","lastName":"Sevarac","linkedInUsername":"zoran-sevarac-phd-49a9a411","mastodonUsername":"","twitterHandle":""}],"summary":"This session shows how modern Java, using the Foreign Function & Memory API and Vector API, can run high-performance AI natively on CPUs and GPUs without JNI or Python. It includes Deep Netts CUDA integration, SIMD acceleration, and a fraud detection use case.","timeSlots":[],"title":"High-Performance Java-Native AI with CUDA and the Vector API","totalFavourites":3,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"For the last nearly two decades, I’ve joined and consulted software teams that all struggled with the same problem: nobody likes writing documentation, and everybody pays the price for it. New developers spend weeks reverse-engineering codebases, architecture lives in people’s heads, and wikis quietly decay into noise.
With AI now entering the picture, many hope it will fill the gaps. It won’t.
I’ll show the simple ideas that change everything: how to think about documentation in the right levels, how to keep it close to the code, and how to avoid the decay we all know too well. These are the decisions that have to be made before any tool or process can work.
The documentation won’t write itself. But there are ways to make it far easier and far more reliable.
Key take-aways:
• How to separate architecture, API, and code-level documentation
• What “docs as code” means once the slogans are gone
• Why missing context stays missing—with or without AI","fieldGroup":null,"fieldGroupValues":[],"id":22942,"keywords":[{"name":"documentation"},{"name":"architecture"},{"name":"ai"},{"name":"code"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Christian Heitzmann, MSc ETH CSE, is the owner of SimplexaCode AG in Lucerne, Switzerland. His professional certifications include Java, Python, and Spring. He holds teaching diplomas in both Computer Science and Mathematics, as well as a CAS in Machine Learning. With over 20 years of experience in professional software development and over 15 years in teaching and training, he works as a developer, documentation consultant, and trainer, bringing order to software, documentation, and data.","blueskyUsername":"","company":"SimplexaCode AG","countryName":"Switzerland","firstName":"Christian","fullName":"Christian Heitzmann","id":23018,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-60ea3c3b-3bd0-4920-b014-21bb439cbb51.jpg","lastName":"Heitzmann","linkedInUsername":"christian-heitzmann","mastodonUsername":"","twitterHandle":""}],"summary":"The talk argues that AI won’t solve poor documentation. It presents practical principles for keeping docs close to code, separating architecture/API/code-level documentation, and preventing decay. Key takeaway: missing context remains missing unless teams make deliberate documentation decisions first.","timeSlots":[],"title":"Why Documentation Won’t Write Itself—and AI Won’t Save You","totalFavourites":5,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Apache Parquet has become the lingua franca of the modern data ecosystem, powering data lakes and table formats like Apache Iceberg—but for Java, the go-to library parquet-java pulls in Hadoop and a truckload of other dependencies, and its reader is single-threaded. This was bugging me enough to start Hardwood, a brand-new Parquet parser written from scratch in modern Java, applying some of the performance lessons learned from the One Billion Row Challenge. Come and join me for this talk, where we'll look at:
- The internals of the Parquet format and what makes parallelizing its decoding surprisingly tricky
- Techniques for achieving high throughput, such as page-level parallelism, adaptive prefetching, and avoiding auto-boxing in hot loops
- How to use JDK Flight Recorder for identifying performance bottlenecks
- Practical learnings from using AI (specifically, Claude Code) as a coding companion—what works well, where you need to stay sharp, and why "built with AI" doesn't mean "vibe-coded"
Whether you're interested in file formats, Java performance, or getting a realistic take on AI-assisted development, there should be something in here for you.","fieldGroup":null,"fieldGroupValues":[],"id":23522,"keywords":[{"name":"performance"},{"name":"java"},{"name":"ai"},{"name":"parquet"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Gunnar Morling is an open-source software engineer in the Java and data streaming space, currently working as a Technologist at Confluent. Previously, he helped to build a realtime stream processing platform based on Apache Flink and led the Debezium project, a distributed platform for change data capture. He is a Java Champion and has founded multiple open source projects such as Hardwood, kcctl, JfrUnit, and MapStruct. Gunnar is an avid blogger (morling.dev) and has spoken at various conferences like QCon, Java One, and Devoxx. He lives in Hamburg, Germany.","blueskyUsername":"gunnarmorling.dev","company":"Confluent","countryName":"Germany","firstName":"Gunnar","fullName":"Gunnar Morling","id":23036,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GihCBYp1plXFDKfAchzpn7tZkBlhGUg3kD7xvlSjw=s96-c","lastName":"Morling","linkedInUsername":"gunnar-morling","mastodonUsername":"https://mastodon.online/@gunnarmorling","twitterHandle":"gunnarmorling"}],"summary":"A talk about Hardwood, a new modern Java Parquet parser built from scratch to avoid Hadoop dependencies and single-threaded limits. It explores Parquet internals, parallel decoding, high-throughput techniques, JDK Flight Recorder profiling, and lessons from AI-assisted development with Claude Code.","timeSlots":[],"title":"Hardwood: Building a Parquet Parser From Scratch (With a Little Help From AI)","totalFavourites":1,"track":{"id":1254,"name":"Cloud Native, Data & Infrastructure","description":"Kubernetes, platform engineering, vector DBs, model serving, and GPU orchestration.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-server-settings.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Ask the Oracle engineers anything related to your favourite programming language & runtime!","fieldGroup":null,"fieldGroupValues":[],"id":23536,"keywords":[{"name":"engineers"},{"name":"runtime"},{"name":"oracle"},{"name":"programming"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"I work in the Java Platform Group at Oracle designing the Java programming language and virtual machine. I am also a member of the Programming Languages Research Group at Oracle Labs. I am based in Cambridge, UK.
My mission is to keep Java as a leading language and platform for developers around the world. Given my background as a researcher, I am especially keen to find, adapt and transfer ideas and techniques from the research community and get them into the hands of Java programmers. I spend much of my time working on Java’s industry-leading specifications.
I joined Oracle in March 2014. Previously I was a senior researcher at Microsoft Research Cambridge from March 2004 until February 2014. Before that I was a University Lecturer at the University of Cambridge Computer Laboratory and a Fellow, College Lecturer and Director of Studies at St John’s College, Cambridge. I studied for my PhD at the University of Cambridge, and for my BSc at Imperial College, London.","blueskyUsername":"","company":"Oracle","countryName":"United Kingdom","firstName":"Gavin","fullName":"Gavin Bierman","id":21503,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-81f196b1-a89f-4d7a-85d3-5e7e4e171a24.jpg","lastName":"Bierman","linkedInUsername":"gavin-bierman-a0173075","mastodonUsername":"","twitterHandle":"GavinBierman"},{"anonymizedBio":null,"bio":"Maurizio Cimadamore is a compiler architect at Oracle. He is currently working on the language and runtime aspects of many foundational projects that will change the shape of the Java platform, such as Project Valhalla and Project Panama.
In the previous years, he has contributed to the design and implementation of several features of the Java programming language, such as diamond inference, lambda expressions and local variable type-inference. Maurizio holds a PhD in language design and implementation.","blueskyUsername":null,"company":"Oracle","countryName":"Italy","firstName":"Maurizio","fullName":"Maurizio Cimadamore","id":21513,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-66579526-c2c5-46b5-87ed-6bb6f34121af.jpg","lastName":"Cimadamore","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null},{"anonymizedBio":null,"bio":"Per Minborg is a manager in the Java Core Library team at Oracle and is also a member of the Panama project team. He is an inventor and developer with more than 25 years of Java coding experience. He has been a frequent contributor to various open-source projects, is a regular speaker at events, Oracle JavaOne alumnus, and co-author of the publication Modern Java.","blueskyUsername":"pminborg.bsky.social","company":"Oracle","countryName":"Sweden","firstName":"Per-Ake","fullName":"Per-Ake Minborg","id":23047,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-a388bc3b-5180-411a-a4a7-5631e4341892.jpg","lastName":"Minborg","linkedInUsername":"per-%C3%A5ke-minborg-448a623b6/?lipi=urn%3Ali%3Apage%3Ad_flagship3_feed%3B4pm%2B8nsET3GGj6bkmM3geg%3D%3D","mastodonUsername":null,"twitterHandle":null},{"anonymizedBio":null,"bio":"Viktor is a Software Architect in the Java Platform Group at Oracle.
For more than two decades, he has focused on enabling developers to solve problems with maximal productivity and maintainability.
Specializing in developer ergonomics in the realm of concurrent, parallel, and distributed programming; he has contributed to the standard libraries of several major programming languages, been involved in more than a dozen Open Source projects, and has spoken at numerous conferences and universities.","blueskyUsername":"","company":"Oracle","countryName":"Sweden","firstName":"Viktor","fullName":"Viktor Klang","id":23049,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTtdkY-Lks0gsRcvl05pqpW8-GOpvuVndojO99ZxUsSy0rgo=s96-c","lastName":"Klang","linkedInUsername":"","mastodonUsername":"","twitterHandle":""},{"anonymizedBio":null,"bio":"Alan Bateman is an architect in the Java Platform Group at Oracle and a long-time contributor to OpenJDK. He leads the OpenJDK Core Libraries Group and has helped shape the evolution of the Java platform through the design and implementation of many JDK features, including New I/O (NIO), JVMTI and serviceability, the Java Module System, and, more recently, Project Loom.","blueskyUsername":"","company":"Oracle","countryName":"Ireland","firstName":"Alan","fullName":"Alan Bateman","id":28801,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-777f0d6e-3ce7-4670-9962-7f3e5f309f46.jpg","lastName":"Bateman","linkedInUsername":"","mastodonUsername":"","twitterHandle":""}],"summary":"Ask Oracle engineers any questions about your favorite programming language and runtime.","timeSlots":[],"title":"Ask the Architects","totalFavourites":9,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Agentic coding generates code faster than any team can review it, and that speed lets a single questionable pattern spread across a codebase before anyone notices. This talk shows how to stop that drift at the source, with layered guardrails that make architectural violations structurally hard to introduce instead of something you catch too late in review.
Drawn from reviewing hundreds of AI-assisted pull requests: natural-language (NL) repo rules don't hold the line against an LLM. The reliable fix is a deterministic foundation, executable architectural rules acting as fitness functions for AI-generated code, built with ArchUnit, custom linters, and CI gates that fail the build the instant a boundary is crossed. That layer is also what makes NL rules work: once it fences off the cases that must never be wrong, targeted NL rules become precise instead of suggestions an agent ignores. Includes live demos of guardrails catching violations in real time.
Takeaways: why NL rules fail to stop LLMs propagating bad patterns; how to build a deterministic enforcement layer (ArchUnit, custom linters, CI gates) with instant feedback; how that layer makes your NL rules reliable; a practical model for sorting any rule into the right layer.
Audience: software architects, tech leads, and senior engineers maintaining architectural consistency on AI-augmented teams.","fieldGroup":null,"fieldGroupValues":[],"id":5507,"keywords":[{"name":"agentic"},{"name":"guardrails"},{"name":"linters"},{"name":"archunit"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"I am a Domain Tech Lead at INSHUR with more than 12 years of experience working on digital transformation initiatives across fintech, insurtech, and cybersecurity. My work has included modernizing legacy platforms, migrating monoliths to microservices, and building large-scale cloud-native systems.
More recently, I have focused on applying AI to digital transformation, leading the development of agentic AI systems and integrating large language models into operational workflows. I am interested in making AI and complex technology initiatives practical, accessible, and understandable across engineering, operations, and business teams.","blueskyUsername":"","company":"INSHUR","countryName":"United Kingdom","firstName":"Sasha","fullName":"Sasha Podlesniuk","id":5807,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-48d9d8c1-f223-44ec-ba3e-6aee8911b78e.jpg","lastName":"Podlesniuk","linkedInUsername":"sasha-podlesniuk","mastodonUsername":"","twitterHandle":""}],"summary":"This talk explains how to prevent AI-generated code from spreading architectural violations by using layered guardrails. It shows why natural-language rules fail, and how deterministic enforcement with ArchUnit, custom linters, and CI gates can stop bad patterns immediately and make higher-level rules effective.","timeSlots":[],"title":"Stopping AI-Generated Architectural Drift: Deterministic Guardrails for Agentic Coding","totalFavourites":5,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Hand two teams the same model and you get wildly different agents. That gap, the system around the model, is where the enterprise questions live: Is it working? How would I know? Can I make it cheaper? Am I locked into one vendor? None is settled by a better model.
This talk introduces Agent Control Theory: treating a coding agent the way engineers treat a drone or autonomous car, an uncertain system you instrument, measure, and steer. Record every run as an immutable journal, judge it against a spec, model the trajectory as a Markov chain, and let the evidence pick the next lever: prompt, skills, workflow, tools, or model, cheapest first.
This is classical control theory, and the math earns its keep: the Markov model yields expected work remaining, and your chosen cost (tokens spent, quality verified) is the performance index the loop steers by. The key is the controller, turning verdict and diagnosis into the next move: human today, AI as the evidence matures.
The agent is not one giant prompt but a workflow of deterministic steps and AI steps, with agentic CLIs (Claude Code, Gemini CLI, Codex) doing the judgment work and Agent Client the portable API to programmatically control them. Once the judges become a benchmark, the agent is a swappable engine: same task, same judges, same evidence. The vendor swap becomes an experiment, not a leap of faith. Source available Java, the AgentWorks stack: agent-workflow, agent-client, agent-journal, agent-judge, agent-experiment.
","fieldGroup":null,"fieldGroupValues":[],"id":25661,"keywords":[{"name":"theory"},{"name":"workflow"},{"name":"control"},{"name":"agent"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Dr. Mark Pollack is the creator of AgentWorks, an open-source agent engineering stack for the JVM, and the founder of Spring AI. After more than twenty years building the Spring ecosystem's data and integration projects, he now works independently on the infrastructure that makes AI agents dependable: durable workflow orchestration, execution tracing and cost metering, evaluation and judging, sandboxed execution, and a multi-provider client that drives agentic CLIs — Claude Code, Gemini, Qwen, Amazon Q — behind a single API. The stack spans agent-workflow, agent-journal, agent-judge, agent-bench, agent-sandbox, agent-hooks, and agent-client, alongside the Agent Client Protocol (ACP) Java SDK and claude-agent-sdk-java, the Java SDK mirroring Anthropic's official SDKs.
His current research, Agent Control Theory, treats coding agents as systems to be measured and steered rather than prompted and hoped for: capture the full execution trajectory, judge the outcome, and use the data to characterize and then improve agent behavior. The approach draws on his pre-Spring career in experimental physics, where he built offline computing systems for high-energy nuclear physics at Brookhaven National Laboratory before moving to front-office trading systems in finance.
Mark has been part of Spring since 2003, when he contributed its JMS support. As founder and lead of Spring AI he brought LLM integration, RAG, vector stores, and tool calling to Java and Spring developers. He previously founded and led Spring Data, Spring Cloud Data Flow, Spring XD, Spring AMQP, Spring Shell, and Spring.NET.
He is a returning speaker at Devoxx Belgium, SpringOne, Spring I/O, and DevNexus, and has appeared on Software Engineering Daily. He builds in the open at lab.pollack.ai and writes at pollack.ai.","blueskyUsername":"","company":"Tuvium","countryName":"United States of America","firstName":"Mark","fullName":"Mark Pollack","id":25502,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-66cdea20-f3d9-49aa-8899-6579e673b1c3.jpg","lastName":"Pollack","linkedInUsername":"marklpollack","mastodonUsername":"","twitterHandle":""}],"summary":"The talk presents “Agent Control Theory” for enterprise coding agents: treat agents as measurable systems, log every run, judge against specs, and use evidence to optimize prompts, tools, workflows, or models. It aims to reduce cost, improve quality, and enable vendor portability through controlled experimentation.","timeSlots":[],"title":" Introducing Agent Control Theory: The System Runs the Model","totalFavourites":7,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Processing large amounts of data efficiently is a common challenge in modern systems. Many workloads spend most of their time performing simple mathematical operations on large arrays of numbers. Modern CPUs can speed this up using SIMD instructions, which allow a single instruction to process multiple values at once. Traditionally, accessing these capabilities from Java has been difficult and often required platform-specific native code.
The Panama Vector API changes this by allowing Java developers to express these operations directly in Java while the JVM turns them into efficient CPU instructions.
Elasticsearch provides an interesting real-world example. Modern features such as vector search rely on large numbers of vector calculations, and analytical queries written in ES|QL often process large batches of numeric data. In both cases the system repeatedly performs simple operations over large arrays of values. Elasticsearch and its underlying search engine Lucene use the Vector API to speed up these operations by taking advantage of SIMD instructions available on modern CPUs.
Using an incubating API in production software is unusual and normally avoided. In this case, however, the potential performance gains are significant enough to make it worthwhile. This talk explores how Elasticsearch integrates the Vector API, how it safely enables and disables SIMD acceleration at runtime, and what this tells us about the future of high-performance computing in Java.","fieldGroup":null,"fieldGroupValues":[],"id":17556,"keywords":[{"name":"elasticsearch"},{"name":"simd"},{"name":"java"},{"name":"vector"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Notorious engineer at work and after hours, tracing meanders of the art of software engineering. Remote Software Gardener, mostly working in web-oriented Java gardens. Java Champion. Docker Captain. Testcontainers Champion. Programming usually in Java (since 1.3), Scala and Go, but in other languages too. Fan of agility, seen mostly as choosing the right tools and approaches after asking the right questions. Developer, trainer and conference speaker. In his talks, Piotr covers not only hardcore Java but also software architecture, computer security, and soft-skills.","blueskyUsername":"piotrprz.bsky.social","company":"Elastic","countryName":"Poland","firstName":"Piotr","fullName":"Piotr Przybyl","id":16384,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-378903c1-6e22-40fd-89cd-46195162ee5a.jpg","lastName":"Przybyl","linkedInUsername":"piotrprzybyl","mastodonUsername":"https://mstdn.social/@piotrprz","twitterHandle":"piotrprz"}],"summary":"This talk explains how Elasticsearch uses Java’s Panama Vector API to accelerate vector search and analytical queries with SIMD instructions. It shows how Lucene and Elasticsearch enable runtime SIMD safely, despite the API being incubating, and discusses implications for high-performance Java computing.","timeSlots":[],"title":"SIMD for Java: how Elasticsearch already benefits from the Panama Vector API","totalFavourites":5,"track":{"id":1254,"name":"Cloud Native, Data & Infrastructure","description":"Kubernetes, platform engineering, vector DBs, model serving, and GPU orchestration.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-server-settings.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Java’s code reflection API, developed in OpenJDK Project Babylon, opens up a new path for targeting foreign programming models directly from Java, including GPUs. This talk introduces the Heterogeneous Accelerator Toolkit (HAT), a Java parallel programming framework to accelerate modern workloads on specialised hardware through its three core stages: reflect, transform, and accelerate.
The reflect phase introduces the basics of code reflection and its key concepts, making the session accessible to developers new to the API. We then show how HAT reflects Java compute code into an analysable code model, transforms that model into a GPU-friendly representation, and accelerates execution by offloading selected computations to OpenCL or CUDA.
Then, this talk shows a set of examples including tensor-core programming and matrix multiplication, fundamental workloads for AI and natural-language processing. The talk also covers performance analysis and tuning of the CUDA backend with NVIDIA Nsight Compute and Nsight Systems, from profiling generated kernels to identifying optimisation opportunities.","fieldGroup":null,"fieldGroupValues":[],"id":21657,"keywords":[{"name":"cuda"},{"name":"reflection"},{"name":"parallelism"},{"name":"gpu"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Juan Fumero, PhD, is a Software Developer in Oracle’s Java Platform Group, specializing in JIT compilation, managed runtime systems, GPU programming, and heterogeneous computing. Previously a Research Fellow at the University of Manchester, he led the development of TornadoVM, a framework for accelerating Java applications on GPUs, FPGAs, and other hardware. He holds a PhD from the University of Edinburgh and has contributed to projects with Intel, ESA, CERN, and Oracle Labs.","blueskyUsername":"jfumero.dev","company":"Java Platrform Group - Oracle","countryName":"Italy","firstName":"Juan","fullName":"Juan Fumero","id":21504,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght9T258iuEoSietb6uF8idCUEZukx0rIGQMYGSWQ=s96-c","lastName":"Fumero","linkedInUsername":"juanjosefumeroalfonso","mastodonUsername":"","twitterHandle":"snatverk"}],"summary":"The talk presents HAT, a toolkit using Java Code Reflection to accelerate Java workloads on GPUs through reflection, transformation, and offloading to OpenCL or CUDA. It includes tensor-core and matrix-multiplication examples, plus performance profiling and tuning with NVIDIA Nsight Compute and Nsight Systems.","timeSlots":[],"title":"Reflect, Transform, Accelerate: Bringing GPU Computing to Java","totalFavourites":2,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Modern software delivery pipelines face increasing pressure from auditors, regulators, and internal governance teams. Yet many builds remain non‑reproducible. This lack of determinism creates blind spots in compliance, undermines traceability, and exposes organisations to risks when validating what actually went into production.
This session demonstrates how developers can use Apache Maven features to reliably produce identical artifacts from identical inputs. You’ll also learn how to diagnose where reproducibility breaks down and how Maven’s built‑in verification tools can highlight sources of non‑determinism.
Join this talk to understand why reproducibility is becoming a non‑negotiable requirement for responsible software development and what you can do to make your build deterministic by design.","fieldGroup":null,"fieldGroupValues":[],"id":4096,"keywords":[{"name":"maven"},{"name":"compliance"},{"name":"determinism"},{"name":"reproducibility"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Maarten is a solution architect, senior Java developer, and technical trainer with a passion for developer productivity and software quality. He is recognised as a Java Champion and Oracle ACE Pro for his contributions to the Java community through speaking, mentoring, and content creation. Maarten is passionate about "building the right thing" and "building the thing right". He loves to share new ideas and knowledge through speaking and his blog. When he's not writing code or speaking at conferences, Maarten enjoys exploring creativity through photography, music, and experimenting in the kitchen - in no particular order.","blueskyUsername":"mthmulders.bsky.social","company":"Info Support","countryName":"Netherlands","firstName":"Maarten","fullName":"Maarten Mulders","id":3927,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-caf6823d-09ef-45bb-8e40-43a6fd1d1e8e.jpg","lastName":"Mulders","linkedInUsername":"mthmulders","mastodonUsername":"https://mastodon.online/@mthmulders","twitterHandle":"mthmulders"}],"summary":"This session shows how Apache Maven can make builds reproducible by producing identical artifacts from identical inputs. It explains how to diagnose non-determinism, use Maven’s verification tools, and improve traceability and compliance in software delivery pipelines increasingly scrutinized by auditors and regulators.","timeSlots":[],"title":"Deterministic by Design - Achieving Reproducible Builds with Maven","totalFavourites":4,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Detailed Description
Temporal makes it remarkably easy to implement long-running business workflows. But once those workflows span multiple services and require synchronous business decisions, straightforward Saga implementations quickly expose non-obvious problems.
We'll explore the challenges we encountered while building distributed transactions with Temporal in a real production system. Starting with a classic Saga implementation, we'll examine why it wasn't enough, what problems emerged around synchronous workflows, failures, cancellation and compensation, and how the architecture evolved to address them. We'll also discuss reconciliation, consistency guarantees, and the trade-offs between keeping workflows simple and building resilient distributed systems.
Key Takeaways
• Common pitfalls when implementing Saga with Temporal.
• Designing reliable synchronous workflows.
Target Audience
Backend engineers, software architects, and tech leads building distributed systems or using Temporal for business workflow orchestration.","fieldGroup":null,"fieldGroupValues":[],"id":16418,"keywords":[{"name":"saga"},{"name":"distributedtransactions"},{"name":"consistency"},{"name":"temporal"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Andrei Sulyz is a Technical Lead with 10 years of experience building distributed systems across startups and enterprise environments. He currently works on the processing platform of a fintech company, designing backend systems for financial transactions and core platform services. His focus is distributed systems, backend architecture and developer productivity. Andrei has led the adoption of Temporal from initial evaluation to production and is now building AI agents that automate software development workflows. He enjoys understanding how technologies work under the hood and sharing practical lessons from real production experience.","blueskyUsername":"","company":"Skyro","countryName":"Spain","firstName":"Andrei","fullName":"Andrei Sulyz","id":16359,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-10dc359e-517c-4cfb-b859-7d7c9ca845a9.jpg","lastName":"Sulyz","linkedInUsername":"andrey-sulyz","mastodonUsername":"","twitterHandle":""}],"summary":"This talk examines real-world challenges of implementing distributed transactions with Temporal. It covers pitfalls of classic Saga patterns, especially for synchronous workflows, failures, retries, and compensation, and explains how the architecture evolved to improve reliability, reconciliation, and consistency in production distributed systems.","timeSlots":[],"title":"Solving Distributed Transactions with the Saga Pattern and Temporal","totalFavourites":7,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"AI coding agents feel like magic. It is easy to assume there is something exotic inside, some secret sauce that makes agents reliable. Well, there isn't, the core of every AI coding agent is very simple: send a message to a model, parse tool calls, execute them, feed the results back, repeat. Everything else (context management, permissions, observability, safety guardrails) is engineering layered on top of it that we should be building.
In this deep dive, we build that engine from scratch (in Java!), live on stage. Not to build the best agent, but to understand how the pieces fit together. We point it at a real codebase, and watch what happens. It compiles. Tests pass. And it violates every convention the team agreed on. So we iterate. We add context, constraints, and feedback, and at each step we examine what changed, why it helped, and what it maps to in the tools you already use.
The goal is a mental model. By the end, you will understand the components inside the AI coding tools you use every day, what you can layer on top to get smoother results and safer expectations, and where the honest limits still are, the gap no amount of engineering has closed yet.","fieldGroup":null,"fieldGroupValues":[],"id":4087,"keywords":[{"name":"tool calls"},{"name":"ai agents"},{"name":"context management"},{"name":"safety guardrails"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Ji is a Data Platform Engineer at Xomnia with a background in Computer Science & Engineering from TU Delft and a recently completed MSc in Artificial Intelligence from Utrecht University. With years of experience in software development, he focuses on solving real-world software and data problems, while maintaining a strong interest in AI, education, and emerging digital policy. He is particularly drawn to questions around EU sovereign tech, AI alignment, and how to build systems that are not only capable, but also compliant, trustworthy, and grounded in human values. Driven by the belief that technology, when built thoughtfully, can be a genuine force for good, Ji is passionate about the role AI can play in improving learning experiences and advancing society responsibly.","blueskyUsername":"","company":"Xomnia","countryName":"Netherlands","firstName":"JI","fullName":"JI Darwish","id":46151,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-69dfe225-6442-4759-bebe-064247f07b54.jpg","lastName":"Darwish","linkedInUsername":"ji-darwish","mastodonUsername":"","twitterHandle":""}],"summary":"AI coding agents are fundamentally simple: prompt a model, execute tool calls, feed results back, and repeat. This talk builds such an engine in Java, then iteratively adds context, constraints, and feedback to show how real tools improve reliability, safety, and understanding of their limits.","timeSlots":[],"title":" Harness Engineering: Building the System Around Your AI Coding Agent ","totalFavourites":9,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"In March 2027, in JDK 28 (JVM gods willing!), after more than 12 years of effort, the Valhalla project will deliver its first feature: Value Classes (JEP 401) as a preview feature. Value classes are defined as regular classes with a "value" keyword in front. They are backward compatible with your code written before Java 28, without you having to recompile it.
Value classes bring new optimization opportunities at the JVM level: scalarization on stack and flattening on heap. For that, the Java specification had to be reinterpreted and enhanced, challenging us to reconsider some of the fundamentals on which Java was created 30 years ago. What does == mean? Is a reference always a pointer? Can an instance of a value class be null? Can you observe the transition between the default values and the final values of fields during object construction? And more.
This presentation covers all these questions, showing the trade-offs that need to be made in order to bring value classes to the Java platform.
You will hear about new code patterns, performance, and type safety. And, as value classes are just the first step, we will cover what Valhalla should bring in the future.","fieldGroup":null,"fieldGroupValues":[],"id":15354,"keywords":[{"name":"performance"},{"name":"valhalla"},{"name":"value classes"},{"name":"jvm"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"José is a Java Developer Advocate at Oracle. He holds a PhD in applied mathematics and computer science and is a Java Champion alumnus. He has been an assistant professor at the University Sorbonne Paris Nord for 25 years. He is an active member of the Paris Java User Group, and a disorganizer of the JChateau unconference. He contributes Java learning content on the dev.java website. He publishes "Java Coding Tips", a biweekly series of shorts, and the "JEP Café", a video series, both on the Java YouTube channel.","blueskyUsername":"josepaumard.bsky.social","company":"Oracle","countryName":"France","firstName":"José","fullName":"José Paumard","id":15304,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GhsVIUuTR9Onoo-D8UMW5QId2Yw1V_EUvqdWdNQ0RQ=s96-c","lastName":"Paumard","linkedInUsername":"jos%C3%A9-paumard-2458ba5","mastodonUsername":"","twitterHandle":""},{"anonymizedBio":null,"bio":"Tenured associate professor at Université Gustave Eiffel (Paris) and longtime OpenJDK member researching runtime environments (Java VM, .NET). He has contributed to major Java features including invokedynamic (JSR 292), Lambda (JSR 335), and the Jigsaw module system (JSR 376), and currently works on OpenJDK's Amber (records, pattern matching) and Valhalla (value types) projects. He's also the main contributor to the ASM bytecode library.
A frequent conference speaker (Devoxx France, FOSDEM), he regularly presents on modern Java features like virtual threads, structured concurrency, and value types.","blueskyUsername":"","company":"University Gustave Eiffel","countryName":"France","firstName":"Remi","fullName":"Remi Forax","id":18075,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocK8ScVAfwuF-N0FTBQLG4RDPeS7vzgzVc_VNU0pEJrFPgQ32w=s96-c","lastName":"Forax","linkedInUsername":"","mastodonUsername":"","twitterHandle":""}],"summary":"Java 28’s Valhalla project introduces preview value classes, enabling JVM optimizations like stack scalarization and heap flattening. The talk explores compatibility, revised Java semantics, nullability, construction behavior, performance, type safety, and new coding patterns, while outlining future Valhalla enhancements.","timeSlots":[],"title":"Will Valhalla Fix Tony's Billion-Dollar Mistake?","totalFavourites":13,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Architects in real life study the monuments built by earlier generations like Cologne Cathedral, Big Ben or Eiffel Tower to understand what makes structures endure. What are the monuments of software to inspire us software engineers? And what can we learn from them for our own software architecture projects?
In this talk we explore a curated set of the most influential software monuments that shaped computing and the architectural ideas behind their lasting impact. From Linux, which grew from a student project into the backbone of modern computing, to the Apollo Guidance Computer that guided humans to the Moon. From Java, one of the most influential programming platforms of the last three decades, to Deep Blue, the machine that defeated the human world chess champion and pushed the limits of computational intelligence.
We revisit the birth of the World Wide Web, the software that started the internet revolution and the simple architectural principles that enabled a global platform for applications and information. And we look at ELIZA, the 1960s chatbot that already revealed both the promise and the surprising psychological effects of human–computer interaction.
Through these stories we uncover the architectural principles that made these systems successful. Principles that still shape modern software systems today. The talk connects history with practice, offering concrete insights and inspiration for engineers designing the next generation of software systems.","fieldGroup":null,"fieldGroupValues":[],"id":12753,"keywords":[{"name":"java"},{"name":"architecture"},{"name":"linux"},{"name":"internet"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Tobias works as a modernization architect at viadee consulting. He supports customers in the modernization of their software architecture and the migration of software systems. His focus sector is insurance and banking. He loves to share his knowledge and experience at international conferences on a regularly basis. ","blueskyUsername":null,"company":"viadee Consultancy","countryName":"Germany","firstName":"Tobias","fullName":"Tobias Voss","id":12607,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-6dacf797-a95e-45a7-885f-141c52896630.jpg","lastName":"Voss","linkedInUsername":"tobiaslvoss","mastodonUsername":null,"twitterHandle":null}],"summary":"The talk explores landmark software systems—Linux, Apollo Guidance Computer, Java, Deep Blue, the World Wide Web, and ELIZA—to reveal enduring architectural principles behind their success and show how these lessons can inspire modern software design.","timeSlots":[],"title":"Software Monuments - Architectural Lessons from the Systems That Shaped Computing","totalFavourites":5,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Decisions, decisions, decision, that is our life every day. But, what is worse than making a bad decision is not remembering why we made a decision, good or bad. Architectural Decision Records are highly important to document the decisions we make, to support the decisions, to bring in necessary and sufficient reasons, and to reevaluate decisions when the business requirements and/or the technologies change. In this presentation we will look at ADRs, how to create them, how to maintain them, and the values they provide.","fieldGroup":null,"fieldGroupValues":[],"id":49256,"keywords":[{"name":"documentation"},{"name":"architectural"},{"name":"decisions"},{"name":"records"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Dr. Venkat Subramaniam is an award-winning author, founder of Agile Developer, Inc., and an instructional professor at the University of Houston.
He has trained and mentored thousands of software developers in the US, Canada, Europe, and Asia, and is a regularly-invited speaker at several international conferences. Venkat helps his clients effectively apply and succeed with agile practices on their software projects.
Venkat is a (co)author of multiple books, including the 2007 Jolt Productivity award winning book Practices of an Agile Developer. His latest book is Functional Programming in Java: Harnessing the Power of Java 8 Lambda Expressions. You can reach him by email at venkats@agiledeveloper.com or on twitter at @venkat_s.","blueskyUsername":null,"company":null,"countryName":"United States of America","firstName":"Venkat","fullName":"Venkat Subramaniam","id":49109,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-83e2efa2-69c8-4599-8ec1-5ec974dd3673.jpg","lastName":"Subramaniam","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null}],"summary":"This presentation explains Architectural Decision Records (ADRs), showing how to create and maintain them and why they matter. It highlights their role in documenting decisions, capturing the reasons behind them, and helping teams reassess choices as business needs or technologies change.","timeSlots":[],"title":"ADRs: The Why and How","totalFavourites":13,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"I've always wanted to create a developer UI for Spring Boot, but the task was massive. The project I ended up doing has 83,000 lines of code, 116 test suites... The honest by-hand estimate for one experienced developer: 6.5 to 8.5 months.
I did it in about 11 days, shipping version 1 after less than 2 weeks.
Not by typing faster: I didn't even open my IDE. I did this by managing a fleet of AI agents: 223 merged pull requests, around 20 a day, with the agents doing all the scaffolding, the panels and the tests while I architected, reviewed and steered. That's roughly a 12–17× reduction in human effort, as measured from my git history and PR metadata.
This talk is about showing you the recipe for delivering at this speed, using modern AI tools. I'll show you how to run it: write the specifications, the test harness, running agents in parallel... And working differently: prep the agents in the afternoon, let them build overnight, merge the results over coffee the next day.
And as a bonus, we get a great new Spring Boot developer console!","fieldGroup":null,"fieldGroupValues":[],"id":11556,"keywords":[{"name":"developer"},{"name":"ai"},{"name":"spring boot"},{"name":"automation"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Julien Dubois is a Java Champion and a Principal Manager at Microsoft. With his team of world-class developer advocates, Julien works directly with the engineering groups to improve how developers can leverage AI using Microsoft and GitHub solutions.
With over 25 years of experience in the Java ecosystem, Julien is best known as the creator of JHipster, an Open Source platform with 22,000+ GitHub stars that revolutionized how developers build Spring Boot applications. His recent work focuses on bringing enterprise-grade AI capabilities to Java through his contributions to LangChain4j and Spring AI.
Julien is a recognized thought leader in the Java community, having spoken at over 200 international conferences including Devoxx, Spring One, and Microsoft Build. He authored Spring par la pratique, the first French book on the Spring Framework, which became a bestseller. His presentations typically feature live-coding sessions and practical workshops that help developers immediately apply new technologies.","blueskyUsername":"","company":"GitHub","countryName":"France","firstName":"Julien","fullName":"Julien Dubois","id":12008,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-832ab0d0-7760-423f-a78e-7f767c7e95b8.jpg","lastName":"Dubois","linkedInUsername":"juliendubois","mastodonUsername":"","twitterHandle":"juliendubois"}],"summary":"An AI-agent workflow enabled building an 83,000-line Spring Boot developer UI in 11 days instead of 6.5–8.5 months. The talk explains how to orchestrate multiple agents with specs, tests, and parallel overnight builds to achieve major productivity gains while delivering a new developer console.","timeSlots":[],"title":"223 Pull Requests in 11 Days: The Recipe for Building Real Software with AI Agents","totalFavourites":28,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Are you tired of juggling dozens of passwords? In 2022 alone, hackers exposed over 24 billion passwords, compromising personal identities, finances, and more. But what if there was a better way? Enter Passkeys. This cutting-edge solution not only offers robust multi-factor authentication but is now paving the way to replace passwords altogether. Join Daniel Garnier Moiroux and Josh Long as they unveil the power of Passkeys. Witness firsthand how the new Spring Security 7 can revolutionize your systems and services, making them more secure and easier to use. Say goodbye to password fatigue and hello to a secure, seamless login experience. Are you ready to embrace the future of authentication?
","fieldGroup":null,"fieldGroupValues":[],"id":14459,"keywords":[{"name":"passwords"},{"name":"authentication"},{"name":"spring security"},{"name":"passkeys"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Daniel Garnier is a software engineer in the Spring team, working on Spring AI, the Java MCP SDK, and Spring Security. He is an adjunct professor at Mines Paris, where he teaches CS and software engineering classes.
He contributes to Spring Security, and has a keen interest in automation and developer productivity","blueskyUsername":"garnier.wf","company":"Spring","countryName":"France","firstName":"Daniel","fullName":"Daniel Garnier-Moiroux","id":7619,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-3e998da2-a249-46ae-baa9-dd880478c765.jpg","lastName":"Garnier-Moiroux","linkedInUsername":"garniermoiroux","mastodonUsername":null,"twitterHandle":null},{"anonymizedBio":null,"bio":"Josh (@starbuxman) has been the first Spring Developer Advocate since 2010. Josh is a Java Champion, author of 7 books (including "Reactive Spring") and numerous best-selling video training (including "Building Microservices with Spring Boot Livelessons"with Spring Boot co-founder Phil Webb), and an open-source contributor (Spring Boot, Spring Integration, Axon, Spring Cloud, Activiti, Vaadin, etc), a Youtuber (Coffee + Software with Josh Long as well as my Spring Tips series ), and a podcaster ("A Bootiful Podcast").
","blueskyUsername":"starbuxman.bsky.social","company":"the Spring team","countryName":"United States of America","firstName":"Josh","fullName":"Josh Long","id":14654,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GhTAlrBfIHBoWpP1mfcKm-kRON4bdBF_LccGehD=s96-c","lastName":"Long","linkedInUsername":"joshlong","mastodonUsername":"","twitterHandle":"starbuxman"}],"summary":"Passkeys offer a passwordless future with stronger, simpler authentication. This talk by Daniel Garnier Moiroux and Josh Long explores how Spring Security 7 can help replace passwords, improve security, and streamline login experiences for users and systems.","timeSlots":[],"title":"Bootiful Security","totalFavourites":18,"track":{"id":1255,"name":"Security, Trust & Compliance","description":"Guardrails, prompt injection defence, model governance, and supply-chain security.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-surveillance-camera.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Coding agents look like magic. Strip away the hype and you find a surprisingly small core: a loop, a language model, and… a shell.
In this Tools in Action, we will see this core live. We give a small, local LLM (under 10B, running on Docker Model Runner) a single built-in tool (bash) and watch it turn into a coding helper. It lists files, greps the repo, runs the tests, curls an API, pipes into jq… all on its own, because the model already knows these CLIs. And when it doesn't? It runs --help and figures it.
We'll see the smallest agent loop in action: the model decides which command to run, the agents executes it, feeds the result back, and repeat until it answers. We'll look at what actually goes over the wire (the tool description, the tool_calls the model emits), why writing a good description is really prompt engineering, and why returning errors is what lets the model recover on its own.
You'll leave understanding how coding agents really work under the hood, why one well-chosen built-in tool often beats a pile of MCP servers, and how far a tiny local model can go.
Live demos, few slides, code you can run tonight.
","fieldGroup":null,"fieldGroupValues":[],"id":22911,"keywords":[{"name":"prompt engineering"},{"name":"bash"},{"name":"coding agents"},{"name":"language model"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Principal Solutions Architect at Docker
- Principal Solution Architect chez Docker
- WebAssembly lover
- GenAI on Raspberry Pi addict
- Container-oriented","blueskyUsername":"k33gorg.bsky.social","company":"Docker","countryName":"France","firstName":"Philippe","fullName":"Philippe Charrière","id":18086,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-c958a486-4d64-44ef-bb6d-77de06638b5d.jpg","lastName":"Charrière","linkedInUsername":"phcharriere","mastodonUsername":"","twitterHandle":"k33g_org"}],"summary":"A live demo shows how a tiny local LLM becomes a coding agent using just a bash tool and a simple loop. It explores tool calls, prompt engineering, error recovery, and why one built-in tool can outperform many servers—revealing how coding agents work under the hood.","timeSlots":[],"title":"One tool to rule them all: bash, a tiny LLM, and the birth of a coding agent","totalFavourites":10,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"Ever wondered how you would write code if you were a seasoned hacker?
In this hands-on session, you’ll learn how to adopt that mindset to build secure Java applications with confidence in the AI era. Every year, thousands of organizations are breached, and billions of records are exposed. Only in 2024, cyberattacks compromised 1.7 billion individuals, and millions of applications were targeted daily. Through "live demos" of real-world attacks and defenses, we’ll cover the essentials every modern Java developer needs in 2026.
Key takeaways:
- A quick look at the OWASP Top 10 (2025) and practical solutions for the newest risks.
- How to implement supply chain security strategies to defend against threats like Shai-Hulud or Log4Shell.
- Ways to integrate essential security checks (SAST/SCA) into your CI/CD pipeline for continuous protection.
- Practical threat modeling techniques to identify and prioritize risks early in the design phase and later in incident response.
- How to set up security monitoring and logging to detect and respond to attacks in production.
- Practical strategies to use AI assistants securely in your development workflow.
- And finally, why using hardened container images is a simple yet highly effective strategy for runtime security.
","fieldGroup":null,"fieldGroupValues":[],"id":8194,"keywords":[{"name":"owasp"},{"name":"java"},{"name":"security"},{"name":"devsecops"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"Soroosh is a software engineer and software architecture enthusiast, passionate about building simple but impactful solutions. With over 15 years of experience in diverse domains, including Telecom, Media & Entertainment, and E-Commerce. He has worked with small startups and large enterprises serving 80 million active subscribers. Currently working as a Solution Architect at Rabobank. Soroosh strongly believes in the power of collaborative learning and enjoys sharing his experiences and insights with other developers.","blueskyUsername":null,"company":"Rabobank","countryName":"Netherlands","firstName":"Soroosh","fullName":"Soroosh Khodami","id":8669,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTtdOw1DaVsLzC-GyXIj6Db1K0D66F-cMGH5gQp9I0Q=s96-c","lastName":"Khodami","linkedInUsername":"sorooshkhodami","mastodonUsername":null,"twitterHandle":"sorooshkh"}],"summary":"Hands-on session for Java developers on building secure applications in the AI era. Covers OWASP Top 10, supply chain security, SAST/SCA in CI/CD, threat modeling, monitoring and logging, secure use of AI assistants, and hardened container images, with live demos of real-world attacks and defenses.","timeSlots":[],"title":"Write Java Code Like a Seasoned Hacker: 2026 Edition","totalFavourites":17,"track":{"id":1255,"name":"Security, Trust & Compliance","description":"Guardrails, prompt injection defence, model governance, and supply-chain security.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-surveillance-camera.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"Modern software increasingly interacts with AI models, query languages, and other specialized programming models, forcing you to switch contexts, embed foreign code, or write verbose wrappers to bridge the gap. What if you could express logic from outside the JVM directly in Java?
In this session, we start by looking at how Java enables you to safely interact with runtimes outside of the JVM using the Foreign Function and Memory API and tools like jextract.
Then will push further with Project Babylon’s code reflection: express AI models as Java programs that Babylon can analyze, transform and lower to inference engines, thus leveraging hardware accelerator backends while skipping external model files or the need to glue code.
Along the way, you will learn how to build expressive FFM-based inference today and author pure Java AI-ready models with Code Reflection tomorrow!","fieldGroup":null,"fieldGroupValues":[],"id":22916,"keywords":[{"name":"java"},{"name":"inference"},{"name":"foreignfunctionmemory"},{"name":"codereflection"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Ana is a Java Champion Alumni and Senior Developer Advocate for the Java Platform Group at Oracle, with an extensive background in Java and cloud development. More recently, her work has expanded to explore how machine learning and AI can be applied within the Java ecosystem, helping Java developers adopt emerging capabilities while continuing to benefit from the stability, tooling, and performance of the Java Platform. She actively supports the growth of technical communities through knowledge sharing and enjoys curating content for conferences as a program committee member. ","blueskyUsername":"ammbra1508.bsky.social","company":"Oracle","countryName":"Netherlands","firstName":"Ana-Maria","fullName":"Ana-Maria Mihalceanu","id":5830,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14Gj1OS6OpAzuwBKwoUuCi9gxvNPvdmVA7nbW8lg3mg=s96-c","lastName":"Mihalceanu","linkedInUsername":"","mastodonUsername":"https://mastodon.social/@ammbra1508","twitterHandle":"ammbra1508"},{"anonymizedBio":null,"bio":"Lize Raes loves helping developers to bring AI into real-world applications. As Developer Advocate for Java + AI at Oracle and collaborator at LangChain4j, she’s invested into making cutting-edge tech accessible and useful for developers. Her path has taken her from cochlear implant research to advising the Belgian government during COVID-19, to bioinformatics for drug development. Lize likes her projects hands-on and slightly unconventional, especially when they help make the world a bit better. Outside of work, you’ll find her behind the piano or in her woodworking atelier.","blueskyUsername":"","company":"Java Platrform Group - Oracle","countryName":"Switzerland","firstName":"Lize","fullName":"Lize Raes","id":18066,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-870e8938-ae57-4b4b-a4e8-db0c8f0e2864.jpg","lastName":"Raes","linkedInUsername":"lize-raes-a8a34110","mastodonUsername":"","twitterHandle":"lizeraes"}],"summary":"The session explores using Java to bridge external runtimes and AI models without glue code. It covers the Foreign Function and Memory API and jextract for safe native interaction, then Project Babylon’s code reflection to express, transform, and lower AI models directly in Java for accelerator-backed inference.","timeSlots":[],"title":"Java and the Next Generation of Programming Models","totalFavourites":16,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Vector search is not hard to prototype: turn data into embeddings, store them, and return the closest matches. But real-world applications need to consider much more: application data, business logic, and practical constraints.
In this session, we’ll go through building a travel application with Micronaut, LangChain4j, and GraalVM.
Some of the topics we’ll cover:
– The strategy for generating vector embeddings
– Combining similarity results with application data and business constraints
– Combining semantic and location-aware search
– Most helpful features of Micronaut Data and Micronaut LangChain4j
– Building and running the application as a Native Image
We’ll see how these pieces work together to turn basic similarity search into a more intelligent and useful application.","fieldGroup":null,"fieldGroupValues":[],"id":24227,"keywords":[{"name":"micronaut"},{"name":"graalvm"},{"name":"vector search"},{"name":"embeddings"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Alina is a developer advocate for GraalVM at Oracle. Loves both programming and natural languages, compilers and performance, and open source. Ambassador of snacks and not running.","blueskyUsername":"","company":"Oracle","countryName":"Switzerland","firstName":"Alina","fullName":"Alina Yurenko","id":19772,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-a31c43d1-c596-4e24-af3f-078f54609e6c.jpg","lastName":"Yurenko","linkedInUsername":"alinayurenko","mastodonUsername":"","twitterHandle":"alina_yurenko"}],"summary":"This session shows how to build an e-commerce vector search app with Micronaut, LangChain4j, and GraalVM. It covers embedding generation, blending similarity with business and location data, key Micronaut features, and Native Image deployment to create smarter, practical search.","timeSlots":[],"title":" Expanding Vector Search Beyond Similarity","totalFavourites":6,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"For the last 20-30 years, the software industry has been learning — sometimes the hard way — how to build better software. Practices like TDD, refactoring, pair programming, simple design, collective code ownership, continuous integration, and coding standards didn't appear by magic. They came from pain, failed projects, and the brilliant minds behind XP — Kent Beck, Ward Cunningham, and Ron Jeffries — who asked: "can we do this better?"
And now? Now we have AI. And suddenly, some developers are asking: "do we still need all that?"
In this session, we'll look at 7 XP practices that are genuinely at risk in an AI-assisted world — not because they're wrong, but because AI makes it tempting to skip them. We'll see what happens when teams do skip them (spoiler: it's not pretty 😅), and more importantly, how these practices don't just survive the AI era — they become *more critical than ever*.
AI amplifies everything. Good habits and bad ones. If your team has no tests, no shared standards, no feedback loop — AI will help you build the mess of your life, faster than ever before.
AI is the best stress test XP practices have ever had. Spoiler: they pass 😉","fieldGroup":null,"fieldGroupValues":[],"id":15359,"keywords":[{"name":"refactoring"},{"name":"xp"},{"name":"ai"},{"name":"testing"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Nacho is a software engineer from Barcelona, fan of eXtreme Programming (XP) practices who has been working with Java and other web technologies before the Y2K. He had experience in different roles in the IT world ('guy-for-everything', developer, team lead, project manager and CTO) working in companies in many sectors: energy, e-commerce, startups (some unicorn) technology consultancy and application performance management. Now working as a Principal Software Engineer at Dynatrace developing software for monitoring applications and in 2022 became Java Champion.
Always concerned with questions like "how does it work?" and "how can we make it better?", Nacho loves to share his knowledge with others and learn from them. As a natural progression to this attitude, he founded the Barcelona Java Users Group and co-founded the Barcelona Developers Conference (DevBcn).
When he is not thinking about the next interesting thing to do for the community, he enjoys spending time with his family, playing sports or improving his TDD and other XP skills.","blueskyUsername":"icougil.bsky.social","company":"Dynatrace","countryName":"Spain","firstName":"Nacho","fullName":"Nacho Cougil","id":15309,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-2a8cc803-44ba-4599-9b4a-53ce17c0085d.jpg","lastName":"Cougil","linkedInUsername":"icougil","mastodonUsername":"https://mastodon.social/@icougil","twitterHandle":"icougil"}],"summary":"AI can tempt teams to skip core XP practices like TDD, refactoring, pair programming, and CI. This session shows why these habits are at risk in AI-assisted development, what breaks when they’re ignored, and why they matter even more as AI accelerates both good and bad engineering.","timeSlots":[],"title":"Is AI killing XP practices? (Or are we letting it?)","totalFavourites":12,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"You have already built AI features into your Java application. The model is wrapped in a service, RAG feeds it context, tools are wired, calls are flowing. It works. Then requirements evolve. A single prompt-and-response is no longer enough. You need steps that follow each other, branches based on decisions, retries when things fail, sometimes several actions at once. The question shifts from "how do I call an LLM?" to "how do I keep it under control?".
That is where agentic systems come in. In three hours we start with a single @Agent doing one job and grow it live into a multi-agent system that can analyze, plan, act, and report back. Each step starts naive. We run it, feel the cracks, then reach for the pattern that holds: sequential, loop, parallel, and conditional workflows. AgenticScope and typed keys keep shared state honest. Goal-oriented agents, an LLM-driven supervisor, and a custom Planner take over when a fixed pipeline stops being enough. We add non-AI agents for deterministic work, human-in-the-loop checkpoints before external actions, and observability with the AgentMonitor, because an agent you cannot see is an agent you cannot debug.
You will leave able to look at a problem and tell whether it wants a plain workflow or a real agent, and how much autonomy to hand over before you lose the thread. The goal is not smarter prompts. It is agentic systems easier to reason about and evolve, because the pattern you do not adopt is the one you do not have to debug.
","fieldGroup":null,"fieldGroupValues":[],"id":10020,"keywords":[{"name":"observability"},{"name":"agentic"},{"name":"supervisor"},{"name":"workflow"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"Loïc Magnette is a seasoned software developer with a strong background in consulting. Currently a senior developer at Oniryx, he specializes in Java and Angular, delivering innovative solutions and sharing his knowledge as a speaker. As a co-organizer of the Belgian Java User Group (BeJUG), he fosters connections within the developer community. Outside of tech, Loïc’s passion for wildlife inspires his work and creativity.","blueskyUsername":"lomagnette.bsky.social","company":"Oniryx","countryName":"Belgium","firstName":"Loïc","fullName":"Loïc Magnette","id":5809,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-cda3ea24-0a68-4fe3-a93c-e199a1d8fedf.jpg","lastName":"Magnette","linkedInUsername":"lomagnette","mastodonUsername":"","twitterHandle":"LoMagnette"}],"summary":"This talk shows how to evolve a Java AI app from simple LLM calls into controllable agentic systems. It covers sequential, loop, parallel, and conditional workflows, shared state, supervisors, planners, human checkpoints, and observability—helping you decide when to use workflows versus autonomous agents.","timeSlots":[],"title":"From AI to Agent: A Field Guide to Agentic Patterns in LangChain4j","totalFavourites":9,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"JPA makes database interaction feel almost magical—until your app grinds to a halt. In this talk, we’ll take a tour through the common (and not-so-common) performance pitfalls that turn object-relational mapping into an object-relational mess. From data loading surprises to connection mismanagement, we’ll explore real-world issues using a sample app: the FunFactStore, where fun facts are sold, but bugs are free.
We’ll dig into how seemingly harmless libraries can sometimes do more harm than good, and how tooling can help you spot trouble before your users do. Whether you’re using Spring Data JPA, navigating tricky transaction boundaries, or just wondering why things suddenly got slow, this session offers practical insights—and maybe a few fun facts.","fieldGroup":null,"fieldGroupValues":[],"id":22905,"keywords":[{"name":"jpa"},{"name":"performance"},{"name":"orm"},{"name":"transactions"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Jos gets a smile on his face when he makes other people happy by letting a computer do what they don’t like to do. He has been trying to do so professionally for the past 17 years.
In his spare time, Jos likes to play tennis. He is a trainer/coach for kids playing the typical Dutch sport korfball.","blueskyUsername":"","company":"EasingYou","countryName":"Netherlands","firstName":"Jos","fullName":"Jos Roseboom","id":21540,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-bc6b28d6-203a-4331-8206-f63ac80f7245.jpg","lastName":"Roseboom","linkedInUsername":"jos-roseboom-75508b11","mastodonUsername":"","twitterHandle":""}],"summary":"This talk examines common JPA performance pitfalls, from unexpected data loading and transaction issues to connection mismanagement. Using a sample app, it shows how libraries and tooling can reveal real-world ORM problems and help prevent slowdowns in Spring Data JPA applications.","timeSlots":[],"title":"When ORM Becomes OMG: Performance Pitfalls in JPA and Friends","totalFavourites":5,"track":{"id":1254,"name":"Cloud Native, Data & Infrastructure","description":"Kubernetes, platform engineering, vector DBs, model serving, and GPU orchestration.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-server-settings.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Most agent frameworks ask a language model to decide what to do next. That works, until it doesn't — and when it doesn't, you're left with a non-reproducible trace and no way to explain what happened. Embabel takes the opposite approach.
Created by Rod Johnson — who calls it the most important project since founding Spring — Embabel is a JVM-native agent framework built on Spring AI and Spring Boot. Its core is Goal-Oriented Action Planning (GOAP), an algorithm borrowed from game AI: you declare goals, annotate actions with @Action and @AchievesGoal, and a deterministic, non-LLM planner chains them by pre- and post-conditions to reach the goal. The plan is code-generated, inspectable before it runs, and explainable afterwards. If Spring AI and LangChain4j are the Servlet API of GenAI, Embabel is Spring MVC.
Full presentation details pending. ","fieldGroup":null,"fieldGroupValues":[],"id":49257,"keywords":[{"name":"deterministic"},{"name":"spring"},{"name":"embabel"},{"name":"goap"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Rod Johnson is a developer, author, entrepreneur and advisor. He is the creator of the Spring Framework and Embabel agent framework.","blueskyUsername":null,"company":"Embabel","countryName":"Australia","firstName":"Rod","fullName":"Rod Johnson","id":49107,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-8bd19dbe-7c4f-4457-970d-dcdcaeea60b0.jpg","lastName":"Johnson","linkedInUsername":"johnsonroda","mastodonUsername":null,"twitterHandle":"springrod"}],"summary":"Embabel is a JVM-native agent framework built on Spring AI and Spring Boot that replaces LLM-driven decision-making with deterministic Goal-Oriented Action Planning (GOAP). Developers declare goals and actions; a code-generated planner chains them by conditions, producing inspectable, reproducible, and explainable execution traces.","timeSlots":[],"title":"Embabel Agent v1.5","totalFavourites":6,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Coding agents are dangerous: they can write files (including their own permissions or scripts meant to keep them in check), scan the filesystem, read credentials, and make tool calls that modify databases or exfiltrate data. Any content the agent processes (a pull request, an external code library, or online documentation) can trigger these actions and lead to persisted compromise. The 'Comment and Control' attack demonstrated this, as did the mini-Shai-Hulud attack, which copied its loader into Claude's settings.json, so it re-executed on every session start.
Built-in agent sandboxes are inadequate. Agents are very good at escaping them (see SandboxEscapeBench) but often need not, as they can disable the sandbox themselves, as specified in the Claude bash sandbox documentation.
Can we make coding agents secure? This talk is for developers running coding agents on their machines or in CI/CD. First we map the attack surface: prompt injection variations (e.g. role-play, poem attack) and how to hide them (Unicode smuggling, custom glyphs). Then we cover seven specific threats (supply chain compromise, data exfiltration) and seven defence layers.
We treat the agent itself as untrusted, clamping it down with externally enforced controls it cannot reach or reconfigure (VMs, network firewalls). Attendees will leave with a sandbox taxonomy to evaluate their own setups, a set of ready-to-use tools that score well against it, and a concrete plan to shrink their agents' blast radius.","fieldGroup":null,"fieldGroupValues":[],"id":7011,"keywords":[{"name":"coding agents"},{"name":"sandbox security"},{"name":"prompt injection"},{"name":"data exfiltration"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Richard Gross is an IT archaeologist at MaibornWolff with more than ten years of modernization experience. His focus is on hexagonal architectures, hypermedia APIs, TestDSLs and the expressive and unambiguous modelling of the domain as code. He enjoys mastering TDD, BDD, DDD, decoupled design and even practices that don't include two D's. He also shaped the open source project CodeCharta, which lets even non-developers grasp the quality of their software.","blueskyUsername":"richargh.de","company":"MaibornWolff GmbH","countryName":"Germany","firstName":"Richard","fullName":"Richard Gross","id":5839,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-7f802c4e-9a8f-467c-99d9-7d96c94b9ebc.jpg","lastName":"Gross","linkedInUsername":"richargh","mastodonUsername":"https://mas.to/richargh","twitterHandle":""}],"summary":"Coding agents can be easily compromised through prompt injection, file access, and tool misuse, leading to persistence or data exfiltration. Built-in sandboxes are insufficient. This talk maps key threats, explains seven defense layers, and recommends externally enforced controls like VMs and firewalls to reduce an agent’s blast radius.","timeSlots":[],"title":"Secure Your Coding Agent Like It’s Malware","totalFavourites":11,"track":{"id":1255,"name":"Security, Trust & Compliance","description":"Guardrails, prompt injection defence, model governance, and supply-chain security.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-surveillance-camera.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"
When code misbehaves, coding agents often fall back on an old habit. They add print statements. The method works. It also changes the program before the agent understands the problem.
The debugger provides a cleaner way to find the evidence. Logpoints inspect application state while the program keeps running. They can write custom messages without changing the source. You can even place them inside compiled third-party libraries. Every output line links back to the code that produced it. This includes logpoints and regular loggers as well as print statements.
IntelliJ IDEA makes its debugger tools available to coding agents. The debugger skill covers dozens of tools and settings. Few developers keep all of them in their heads. The complete playbook fits into the agent's context.
The agent combines that debugger knowledge with its understanding of the project code. It can choose useful locations for breakpoints and logpoints. It can start a debug run configuration and read actual values. As the evidence develops, it can move its observation points. The source stays untouched.
What attendees will learn:
- How Logpoints capture runtime state without suspending the program or modifying its source
- How to investigate behavior inside compiled third-party libraries
- How navigable debugger output connects runtime observations to their origin in code
- How an agent combines code context with debugger knowledge to place useful breakpoints and logpoints
","fieldGroup":null,"fieldGroupValues":[],"id":23533,"keywords":[{"name":"debugger"},{"name":"logpoints"},{"name":"runtime"},{"name":"intellij"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Anton is a Developer Advocate at JetBrains, where he focuses on IntelliJ IDEA and the intersection of AI agents and developer tooling. A Java Champion since 2014 with a background in server-side development, he has spent over a decade building tools and content for developers. These days, you'll find him speaking at conferences, exploring agentic workflows, and looking for ways to make the development experience smarter and more enjoyable.","blueskyUsername":"antonarhipov.bsky.social","company":"JetBrains","countryName":"Estonia","firstName":"Anton","fullName":"Anton Arhipov","id":5831,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-d4d89f78-d4bb-456f-ab6f-4768b4822745.jpg","lastName":"Arhipov","linkedInUsername":"antonarhipov","mastodonUsername":"","twitterHandle":"antonarhipov"}],"summary":"IntelliJ IDEA’s debugger tools let coding agents inspect runtime state without altering source code. Using logpoints, agents can trace behavior in running code, including compiled third-party libraries, and link output back to its origin. The debugger skill guides agents to investigate bugs more cleanly than print statements.","timeSlots":[],"title":"Teaching a New Dog Old Tricks: Give the Agent a Debugger","totalFavourites":8,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"We've all been there: you need a quick automation script, so you reach for Bash or Python and immediately get tripped up by syntax you can never remember. Or you try JavaScript and watch 5 + 3 quietly become "53". You want something fast, but you also want type safety and a language you actually know.
Good news: modern Java has become a real scripting language. Shebang support, single-file execution, and instance main methods mean no boilerplate and no build step—just a file you can run. When you need external libraries, JBang handles dependencies with a single line. And unlike a hastily typed shell script, the compiler still has your back before anything runs.
Through live demos, we'll go from a throwaway one-liner to genuinely useful scripts, using records and text blocks to keep them short and readable: parsing a CSV and calling an API, or building a self-contained tool your AI harness can invoke, returning typed, predictable output instead of best-effort text. All in Java, and all without a pom.xml in sight. You'll leave with a new tool for the jobs you currently hand off to Bash.","fieldGroup":null,"fieldGroupValues":[],"id":10021,"keywords":[{"name":"jbang"},{"name":"scripting"},{"name":"java"},{"name":"automation"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Loïc Magnette is a seasoned software developer with a strong background in consulting. Currently a senior developer at Oniryx, he specializes in Java and Angular, delivering innovative solutions and sharing his knowledge as a speaker. As a co-organizer of the Belgian Java User Group (BeJUG), he fosters connections within the developer community. Outside of tech, Loïc’s passion for wildlife inspires his work and creativity.","blueskyUsername":"lomagnette.bsky.social","company":"Oniryx","countryName":"Belgium","firstName":"Loïc","fullName":"Loïc Magnette","id":5809,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-cda3ea24-0a68-4fe3-a93c-e199a1d8fedf.jpg","lastName":"Magnette","linkedInUsername":"lomagnette","mastodonUsername":"","twitterHandle":"LoMagnette"}],"summary":"The talk shows how modern Java has become practical for scripting and automation, thanks to shebang support, single-file execution, and instance main methods. It also covers JBang for dependency management and advanced use cases, with live demos demonstrating fast, maintainable Java scripts.","timeSlots":[],"title":"Java Script (No, Not That One): Modern Automation with Java","totalFavourites":5,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"What do you learn after spending years maintaining and evolving a technology that predates modern programming languages, cloud platforms, and smartphones?
This session explores real engineering challenges encountered while building PDF technology at scale, from compatibility and performance to security, testing, and interoperability. Through a series of stories and hard-earned lessons, you'll see why mature systems can be some of the best teachers for designing resilient software and why the hardest problems are often the ones users never notice.","fieldGroup":null,"fieldGroupValues":[],"id":50104,"keywords":[],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Software engineer at Apryse, specializing in PDF technologies and the iText SDK. With nearly 7 years of experience in backend development, he works on core PDF internals, standards compliance.","blueskyUsername":"","company":"Apryse","countryName":"Belgium","firstName":"Guust","fullName":"Guust Ysebie","id":50054,"imageUrl":null,"lastName":"Ysebie","linkedInUsername":"guust-ysebie-648749291","mastodonUsername":"","twitterHandle":""}],"summary":"A session on engineering lessons from maintaining and scaling legacy PDF technology, covering compatibility, performance, security, testing, and interoperability. It shows how mature systems teach resilient software design and highlights the invisible challenges behind reliable user experiences.","timeSlots":[],"title":"Engineering Lessons from Working on Software Older Than Me","totalFavourites":null,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Have you ever wondered what it takes to become a conference speaker? Where do you start? Which topic do you pick? And how do you actually deliver a talk that lands?
In this session, a group of experienced speakers will answer these questions and more. We'll share tips on getting started, writing proposals that get accepted, preparing your material, and delivering a great presentation. We'll talk about what worked, what didn't, and the lessons we learned along the way. You'll leave with practical takeaways - and the confidence to submit your first proposal.","fieldGroup":null,"fieldGroupValues":[],"id":19707,"keywords":[{"name":"speaking"},{"name":"proposals"},{"name":"presentations"},{"name":"technical events"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":60,"freeTickets":1,"id":954,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"BOF","pause":false,"scheduleInSmallRoom":true,"slug":"bof60"},"speakers":[{"anonymizedBio":null,"bio":"Bert Jan is CTO at OpenValue and focuses on Java, AI, software architecture, Continuous Delivery and DevOps. Bert Jan is a Java Champion, JavaOne Rock Star speaker, Duke's Choice Award winner and leads NLJUG, the Dutch Java User Group. He loves to share his experience by speaking at conferences, writing for the Dutch Java magazine and helping out Devoxx4Kids with teaching kids how to code. Bert Jan is easily reachable on BlueSky via @bjschrijver.dev.","blueskyUsername":"bjschrijver.dev","company":"OpenValue","countryName":"Netherlands","firstName":"Bert","fullName":"Bert Jan Schrijver","id":19753,"imageUrl":"https://avatars.githubusercontent.com/u/4464854?v=4","lastName":"Jan Schrijver","linkedInUsername":"bjschrijver","mastodonUsername":null,"twitterHandle":"bjschrijver"}],"summary":"This session, led by experienced speakers, offers practical guidance on starting technical speaking: choosing topics, writing proposals, preparing talks, and delivering effectively. It shares lessons learned, successes, failures, and key takeaways to help attendees begin their own speaking journey.","timeSlots":[],"title":"Getting started with speaking: Your first conference talk","totalFavourites":8,"track":{"id":1253,"name":"People, Practices & Career","description":"Leading teams, engineering culture, and career resilience.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-deepfake-ai.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"In this session I will share best practices for using MCP setups and AI Agents in real development workflows 🤖. The goal is to improve both quality and security, since even with all the excitement around these tools, it’s still very easy to create confusing interactions, accidental data exposure, or Agents that trigger actions we never intended.
I will also present reports comparing how different LLM models generate code, showing clear differences in stability, output consistency, and the amount of guidance each model needs to avoid mistakes.
And last but not least, I will talk about the elephant in the room: metrics 📏. How do we measure our improvement using AI? Is it something companies are doing right? Spoiler alert ... no 😰
The talk focuses on helping teams build more reliable, productive, and secure AI workflows, making AI code generation a real benefit rather than a risky experiment 😅.","fieldGroup":null,"fieldGroupValues":[],"id":9032,"keywords":[{"name":"security"},{"name":"metrics"},{"name":"agents"},{"name":"mcp"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"International Speaker, JavaChampion, Cofounder of JBCNConf and DevBcn conferences in Barcelona, and AI4Devs conference in Amsterdam.
Working as a Principal AI Development Enabler at GK Software, taking care of the AI Adoption with best practices and tooling and helping the company as the Engineering Excellence Lead.
Formerly Senior Staff Developer Advocate in Java at Sonar (SonarQube), focused on Code Quality, Dev Productivity, AI & Security.
I have worked as a (paid) developer for more than 30 years ago using multiple languajes, but for the last 15 using Java. Although I started when I was 14 with my Amstrad CPC 6128 :)
I am very interested in simulated reality, psychology, philosophy, critical thinking, AI and Java.","blueskyUsername":"","company":"GK Software","countryName":"Spain","firstName":"Jonathan","fullName":"Jonathan Vila López","id":8693,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-f359c583-9652-403e-8cf9-20a2e4265cef.jpg","lastName":"Vila López","linkedInUsername":"jonathanvila","mastodonUsername":"","twitterHandle":"vilojona"}],"summary":"This session shares best practices for MCP setups and AI agents in development workflows, focusing on quality, security, and avoiding unintended actions. It compares LLM code-generation stability and consistency, and addresses how to measure AI-driven improvement, helping teams build reliable, productive, and safe AI workflows.","timeSlots":[],"title":"Agents With Seatbelts: Practical Ways to Keep AI Code Gen Under Control","totalFavourites":22,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Maven is more than 20 years old, and it's last major version (Maven 3) was released in 2010. Since then Maven got several great updates, but the Maven team also realized that Maven can't evolve further without huge changes which have the capability to break the whole ecosystem if done wrong. Now - after years of intensive planning and coding - the release of Maven 4 is closer than ever before!
In this talk I'll speak about the motivation and challenges the team had while developing Maven 4 and of course about its new features and improvements.","fieldGroup":null,"fieldGroupValues":[],"id":21675,"keywords":[{"name":"maven"},{"name":"version"},{"name":"features"},{"name":"ecosystem"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Matthias loves Java, Jakarta EE / Microprofile, testing and documentation. He is a Maven committer and maintainer of the JUnit extension library "JUnit Pioneer".","blueskyUsername":"bukama.bsky.social","company":"ITZBund","countryName":"Germany","firstName":"Matthias","fullName":"Matthias Bünger","id":3924,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-c097beff-4de0-4a4d-ab32-87019a3d1108.jpg","lastName":"Bünger","linkedInUsername":"matthias-bünger-a71b60282","mastodonUsername":"https://mastodon.social/@bukama","twitterHandle":""}],"summary":"This talk explores Maven 4’s long-awaited development, the challenges of evolving a mature ecosystem without breaking compatibility, and the motivations behind its redesign. It also highlights key new features and improvements in the upcoming major release.","timeSlots":[],"title":"All you need to know about Maven 4!","totalFavourites":16,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"We are living through a revolution.
From assisting us in writing code to coordinating a fleet of agents, we find ourselves in the eye of the storm. Coding agents like Claude Code, Copilot, Codex CLI, and Cursor are reshaping our craft and our place within the company.
I brought Claude Code to my company, Back Market, in June 2025. We were "behind" other companies, such as Doctolib. Step by step, we moved from smart autocompletion to writing features end-to-end. Today we have Product Managers who write code, developers who do architecture and design, and managers who build their own internal tools—all with Claude Code. And there's me, looking at how we can control our monthly spending, leading a team of 40 subject-matter experts, with successes (and failures).
But more broadly, what's the flip side? What happens when a junior developer opens a single pull request with 50 modified files using Claude Code? What happens when one team migrates a module in place of the team that originally owned it, with Claude Code? What are the risks, and how do you stay in control within your company?
Finally, as developers—whether junior or senior—this asks us to adapt.
How do we get up to speed? What does prompting actually mean? How does an agent work? What are Skills? What's the impact on CI/CD? On the Platform engineering teams? On the run side? On upfront design?
I don't think AI will replace developers. I think developers who use AI can replace us.","fieldGroup":null,"fieldGroupValues":[],"id":8171,"keywords":[{"name":"coding"},{"name":"developers"},{"name":"ai"},{"name":"agents"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Nicolas is a Principal Engineer at Back Market, a leader in refurbished electronic devices in 18 countries. He is also the co-creator and organizer of Devoxx France. He has a JVM background but has worked in different industries and technologies. Before joining Back Market, he was a Principal Engineer at Doctolib, during the COVID19 crisis.
He's also the author of "le Touilleur Express", a blog about the journey of a software developer.
He has a YouTube Channel with more than 800 subscribers, on GenAI and Coding Agent.
","blueskyUsername":null,"company":"Back Market","countryName":"France","firstName":"Nicolas","fullName":"Nicolas Martignole","id":7648,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-e4f8b008-e24f-4ca0-a520-acbb98ac703a.jpg","lastName":"Martignole","linkedInUsername":"nmartignole","mastodonUsername":null,"twitterHandle":"nmartignole"}],"summary":"AI coding agents are transforming software development, enabling non-developers to build features and teams to work end-to-end. The talk explores real-world adoption, risks like large uncontrolled changes, cost management, and how developers and organizations must adapt to prompting, agents, skills, and CI/CD.","timeSlots":[],"title":"0 to 350 Engineers on Claude Code: What Could Go Wrong?","totalFavourites":21,"track":{"id":1253,"name":"People, Practices & Career","description":"Leading teams, engineering culture, and career resilience.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-deepfake-ai.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"An AI deletes a production database… and insists it didn't. Welcome to modern software engineering.
95% of developers now use AI tools weekly. Claude Code, Cursor, Copilot - agentic coding assistants are writing more code than ever. But the data shows a darker side: up to 5 times more technical debt, over 40% higher code complexity, and a flood of code that needs review.
The real shift isn't about writing more code faster. It's about moving from writing code to writing specs, from building systems to generating them, and from trusting output to verifying it.
In this talk, I'll share real stories - including a production database incident, a 45-minute spec-to-app demo, and lessons from 3 projects built in 1 hour. You'll leave with a practical framework: experiment boldly, deploy carefully, and raise the bar on testing. One rule applies no matter what: if you bring it to a review, it's your code.","fieldGroup":null,"fieldGroupValues":[],"id":19706,"keywords":[{"name":"accountability"},{"name":"softwareengineering"},{"name":"verification"},{"name":"generativeai"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Bert Jan is CTO at OpenValue and focuses on Java, AI, software architecture, Continuous Delivery and DevOps. Bert Jan is a Java Champion, JavaOne Rock Star speaker, Duke's Choice Award winner and leads NLJUG, the Dutch Java User Group. He loves to share his experience by speaking at conferences, writing for the Dutch Java magazine and helping out Devoxx4Kids with teaching kids how to code. Bert Jan is easily reachable on BlueSky via @bjschrijver.dev.","blueskyUsername":"bjschrijver.dev","company":"OpenValue","countryName":"Netherlands","firstName":"Bert","fullName":"Bert Jan Schrijver","id":19753,"imageUrl":"https://avatars.githubusercontent.com/u/4464854?v=4","lastName":"Jan Schrijver","linkedInUsername":"bjschrijver","mastodonUsername":null,"twitterHandle":"bjschrijver"}],"summary":"This talk explores how generative AI is changing software engineering—from coding to verification and guidance. It shares lessons, successes, failures, and responsible practices for using fast-evolving AI tools. The key message: even when AI does the work, humans remain accountable for the outcome.","timeSlots":[],"title":"Software engineering in the age of AI agents: who’s in control?","totalFavourites":20,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"For most of our careers, the scarce thing was producing code. That world is ending. An assistant now produces competent code faster than any of us can read it, and the bottleneck has moved from writing to understanding.
Nothing moved on the responsibility side, though. We are still accountable for every line we merge, no matter who or what typed it. Staying in the loop, knowing the system, owning the design, that is the job description now.
This talk is not a manifesto, it is a set of recipes. The practices I use daily to stay in control of code that arrives faster than I could write it:
- Phased delivery - Large tasks are split into small phases with a review gate after each one. No big-bang diffs.
- Walk-throughs - The assistant guides me through a change step by step, in reading order, instead of leaving me alone with a raw diff.
- Made for humans - Correct is not the bar, understandable is. Design and readability are acceptance criteria, and I send back what does not meet them.
- Engineering depth - Know your stuff. The deeper you understand your system, the harder it is to be fooled by a confident, plausible answer.
- Files and comments - Sessions produce files and in-code comments instead of long chat threads. Easy to navigate, easy to annotate for the assistant.
Together these form a flow that keeps me in control and grows my knowledge instead of eroding it. You will leave with techniques you can apply the next day, whatever assistant you use.","fieldGroup":null,"fieldGroupValues":[],"id":25042,"keywords":[{"name":"readability"},{"name":"understanding"},{"name":"control"},{"name":"code"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Renato Cavalcanti is an independent software engineer and consultant specializing in distributed systems, with deep expertise in Event Sourcing, CQRS, and Domain-Driven Design. He has been a contributor to Akka since 2017, working on the Akka libraries, and most recently Akka Runtime and Akka SDK, where he applies distributed systems principles, such as durability, workflows, supervision, and observability, to the design of reliable multi-agent AI systems.
A Scala developer since 2009, he founded BeScala, the Belgian Scala User Group, and is a regular speaker at Scala, Java, and broader developer conferences, as well as workshops on Scala and Event Sourcing.","blueskyUsername":"renato.cavalcanti.io","company":"Octo Cogito","countryName":"Belgium","firstName":"Renato","fullName":"Renato Guerra Cavalcanti","id":24197,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-c563b2ce-c0f7-426b-8f1f-81169b8efe6b.jpg","lastName":"Guerra Cavalcanti","linkedInUsername":"renatocavalcanti","mastodonUsername":"","twitterHandle":""}],"summary":"As AI writes code faster than humans can read it, the bottleneck shifts to understanding and accountability. This talk shares practical techniques—phased delivery, guided walk-throughs, human-centered review, engineering depth, and file-based collaboration—to keep developers in control, improve comprehension, and maintain design ownership.","timeSlots":[],"title":"Owning the Code You Didn't Type","totalFavourites":17,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"AI is having a huge impact on our software development teams - nobody is sure how the role of a developer will evolve, and Simon and Katie don’t have the perfect prediction either!
Now that our teams have ever more capable coding assistants working around the clock, we are seeing “new” bottlenecks appear around value and trust questions. In response to this, we’re seeing organizations create new flavors of engineering roles, move down to one-pizza-teams, move to product owners and their AI wrangler (a one-pizza-slice team?), or even build factory lines.
This is not the first time the way we approach software teams has changed: we’ve gone from silos and super specialization to cross-functional Scrum teams and 2-pizza teams; from T-shaped and M-shaped professionals to full-stack, full-lifecycle, DevSecFin…Ops, value streams, Team Topologies, and more. We’ve added all kinds of coaches and then removed them again.
So who is invited to the pizza party of the future? Who makes decisions, who takes responsibility, what new and old roles and skills will be needed? And who is no longer welcome? We’ll take a look at what some of the friendlier agile critics from the past can warn us about. We will see how their takes can help us move to a more value-oriented future with agents and humans collaborating on user needs.","fieldGroup":null,"fieldGroupValues":[],"id":8164,"keywords":[{"name":"teams"},{"name":"software"},{"name":"ai"},{"name":"agile"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Katie is a delivery lead, manager, (ex) software developer, and trainer at Reply Comsysto. Over the past 8 years, she has been involved in complex software projects spanning various domains, from semiconductor manufacturing to food supply chains.
Developing quality software products requires a team effort. Therefore, her focus is on fostering a collaborative team culture, promoting an agile mindset, and continuously evolving suitable processes within the organization.","blueskyUsername":null,"company":"Comsysto Reply","countryName":"Germany","firstName":"Katie","fullName":"Katie Clark","id":5803,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-02e30f45-7d0b-455a-9fbf-17e105b17359.jpg","lastName":"Clark","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null},{"anonymizedBio":null,"bio":"Simon Rohrer, Head of Enterprise Architecture and Ways of Working at Saxo Bank has spent his career advocating for modern ways of working, with roles in hands on software engineering, agile coaching with a focus on technical excellence, and senior roles in enterprise architecture and strategy.
He's a co-author of Jonathan Smart's IT Revolution Press book Sooner Safer Happier, and working on a chapter on organising for outcomes for the forthcoming follow-up book.","blueskyUsername":null,"company":"Saxo Bank","countryName":"Denmark","firstName":"Simon","fullName":"Simon Rohrer","id":7602,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-399454c9-6bcf-448e-ab7d-778aac81bcc3.jpg","lastName":"Rohrer","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null}],"summary":"AI is reshaping software teams, creating new bottlenecks around value and trust. The talk explores how developer roles, team structures, and responsibilities may evolve—from cross-functional teams to AI-assisted models—drawing on past agile critiques to imagine a more value-driven future of human-agent collaboration.","timeSlots":[],"title":"The pizza party of the future 🍕 software teams and product value in the age of coding agents","totalFavourites":10,"track":{"id":1253,"name":"People, Practices & Career","description":"Leading teams, engineering culture, and career resilience.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-deepfake-ai.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"“It depends.” It’s the classic software architect’s answer… but what does it actually depend on? Patterns like Clean and Hexagonal Architecture are great tools, but they are often applied dogmatically. When we automatically wrap simple features in endless layers of abstraction and domain interfaces, we introduce unnecessary distance between related concepts. Those extra layers are intended to decouple, but they can sometimes just make our code harder to navigate, test, and maintain. In this talk, we’ll take a step back from “best practices” and look at the trade-offs behind architectural choices. Using clear Java code samples, we will explore the metrics that help guide these decisions: coupling, cohesion, and distance. Instead of looking for universally “clean” solutions, we’ll discuss how to evaluate what makes sense for your specific context.","fieldGroup":null,"fieldGroupValues":[],"id":8186,"keywords":[{"name":"cohesion"},{"name":"architecture"},{"name":"abstraction"},{"name":"coupling"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Ties is a software engineer with a passion for concepts, software engineering fundamentals and helping others. He combines these passions by doing public speaking, volunteer work for organisations like Devoxx4kids and codingcoach and working as a Software Quality Expert at Alliander.","blueskyUsername":"","company":"Alliander","countryName":"Netherlands","firstName":"Ties","fullName":"Ties van de Ven","id":8665,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-0f6945a0-82c6-4715-8549-69f9740c0197.jpg","lastName":"van de Ven","linkedInUsername":"tiesvandeven","mastodonUsername":"","twitterHandle":""}],"summary":"The talk critiques dogmatic use of Clean and Hexagonal Architecture, arguing that unnecessary abstraction can hurt code clarity and maintainability. Using Java examples, it explores trade-offs and decision metrics—coupling, cohesion, and distance—to help choose architecture based on context rather than universal “best practices.”","timeSlots":[],"title":"The Hexagonal Hangover","totalFavourites":16,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"While some folks believe Open Source is important to do Because It's The Right Thing To Do, others have been able to persuade their organisations to open source valuable IP because it brings something to the company. Still others have been able to make money off an open source project they started.
Has this changed in the age of AI? If AI is being trained on open source and freely available code, does that make open source libraries and tools more or less valuable? Can an organisation still make money off their open source tools if AI agents can just go ahead and create code for them without needing training, consulting or support contracts?
Or was the value never in the money it brought, but in the community story? Making your code freely available, creating useful documentation and example code, supporting the community that grows around your open source tool might be the real goal, not simply the path to the destination.
Trisha wants to bring together open source maintainers, contributors and enthusiasts to discuss difficult questions around running an open source tool. Like:
- How does it support itself financially?
- How do you decide what belongs in an open source offering vs what should be commercial/paid?
- Why do we create open source tools in the first place?
- What are the pros and cons of having open source tools in an AI world?
- How do we build businesses around open source tools?
- How do we use open source tools to support businesses?
","fieldGroup":null,"fieldGroupValues":[],"id":8167,"keywords":[{"name":"ai"},{"name":"opensource"},{"name":"community"},{"name":"business"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":60,"freeTickets":1,"id":954,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"BOF","pause":false,"scheduleInSmallRoom":true,"slug":"bof60"},"speakers":[{"anonymizedBio":null,"bio":"Trisha Gee is a Java Champion, author, and internationally recognised speaker who has spent over 20 years building software and helping developers build it better. From high-performance financial systems to developer tools, she brings hands-on experience and a clear, practical perspective on what actually makes teams effective, and how engineering organisations build better software.
She is the co-author of Head First Java and Getting to Know IntelliJ IDEA, and her talks, articles, and videos are widely used by developers to navigate complex topics and improve how they work. She focuses on developer productivity, proven engineering practices, and the role of human judgment in an increasingly AI-assisted world.
Trisha got her start in developer communities, founding and supporting user groups and engaging directly with developers. She continues to champion community-led learning, inclusive practices, and the idea that great software comes from people, not just tools.","blueskyUsername":"trishagee.bsky.social","company":"Digital Aguacates","countryName":"Spain","firstName":"Trisha","fullName":"Trisha Gee","id":7646,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocLsMb8CUpH7W48b3uQ9kMWx6uoYjj07hJVCxmz2MT5fbaMVope5ow=s96-c","lastName":"Gee","linkedInUsername":"trishagee","mastodonUsername":"https://mastodon.social/@trishagee","twitterHandle":"trisha_gee"}],"summary":"This talk explores how AI changes the value of open source, asking whether libraries and tools remain commercially useful when AI can generate code. It brings together maintainers and contributors to discuss funding, community, business models, and the purpose of open source in an AI-driven world.","timeSlots":[],"title":"Open Source in the Age of AI: Where's the value?","totalFavourites":8,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"More and more of us use Coding Assistants that excel at **writing code**.
They are good and they are expensive!
The same underlying models can be used to write
more specialized, cheaper, more predictable AI agents.
+ Maybe you need an agent specialized for your business.
+ Or you need a cheap, disposable one for a quick and dirty task.
+ Later, you want multiple agents to collaborate.
All of that can be done with no code, leveraging Docker Agent,
an open-source project and a real Swiss Army knife for AI agent creation:
+ Create intelligent agents that solve real problems in minutes, without code.
+ Test, benchmark, and optimize agent behavior systematically through evals.
+ Build collaborative multi-agent architectures with MCP, ACP and A2A.
+ Work with any LLM provider on your own terms: frontier, local, small, large.","fieldGroup":null,"fieldGroupValues":[],"id":19713,"keywords":[{"name":"docker agent"},{"name":"ai agents"},{"name":"multi-agent"},{"name":"coding assistants"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Engineer at Docker, working an AI agents.","blueskyUsername":null,"company":"Docker","countryName":"France","firstName":"David","fullName":"David Gageot","id":19756,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-00adb09a-3b25-4aa0-a9c0-c7a771d40845.jpg","lastName":"Gageot","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null}],"summary":"Docker Agent is an open-source, no-code platform for creating specialized AI agents quickly and affordably. It supports testing, benchmarking, optimization, and collaborative multi-agent workflows, while working with any LLM provider.","timeSlots":[],"title":"Create your own AI Agents without code","totalFavourites":11,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"This conference started in 2001 as Javapolis, the same year J2EE's first specs were taking shape. Twenty-five years later, on this stage, we're closing that loop: implementing those specs with a 2026 AI coding agent as co-author, live.
Part 1 — Building Vidocq, then building something new, live. We'll walk through how Vidocq, our Jakarta EE Core Profile 11 and MicroProfile 7.1 runtime, came together: the choice to code against formal specifications rather than our own assumptions, and what changed once AI joined that process. Where AI-generated code passed our review but failed spec conformance. Where a spec's edge cases caught something the AI happily hallucinated past. The workflow we landed on to keep AI output honest against TCK-level requirements. Then we'll do it live: implementing a Jakarta EE or MicroProfile spec we haven't built yet, on stage, with an AI agent constrained by the spec text. Real live coding, backup video included in case it goes sideways.
Part 2 — Vidocq, from spec to CLI. A fast tour of what defines the runtime: JPMS enforced at unit granularity, Build Compatible Extensions, and the JDK 25 features (virtual threads, structured concurrency, scoped values) behind Vidocq's zero external dependencies. Then straight to the command line: we'll bootstrap and build a small app on Vidocq live, showing how those choices feel to work with, not just how they look in a diagram.
Two decades of spec work, one AI model, one CLI. Watch the gap close.
","fieldGroup":null,"fieldGroupValues":[],"id":17765,"keywords":[{"name":"cdi"},{"name":"jpms"},{"name":"microprofile"},{"name":"jakartaee"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"Antoine is a Java Champion and former spec lead for Jakarta EE CDI. He was part of the founding team behind Quarkus at Red Hat and worked on major MicroProfile specifications such as Fault Tolerance and Health Check. Today, Antoine is an architect and technical expert at SCIAM, where he leads complex projects largely built on the very technologies he helped design. He is also the co-creator of the Vidocq framework, a modern MicroProfile stack built on Java 25 and JPMS.","blueskyUsername":"antoine.sabot-durand.net","company":"SCIAM","countryName":"France","firstName":"Antoine","fullName":"Antoine Sabot-Durand","id":18064,"imageUrl":"https://avatars.githubusercontent.com/u/489817?v=4","lastName":"Sabot-Durand","linkedInUsername":"antoinesabotdurand","mastodonUsername":"","twitterHandle":""},{"anonymizedBio":null,"bio":"Yann Blazart is a Senior Tech Lead and a passionate advocate of Jakarta EE and Quarkus. He has been in love with Java for 26 years and has been working with SCIAM since 2019. A long-time open-source contributor, Yann has taken part in projects such as WicketExtJs, TomEE, and JAX-WS Metro.
He is no stranger to the Devoxx stage: he ran two Hands-on Labs at Devoxx France (2012 and 2015) on Java EE, and more recently delivered a Hands-on Lab on Quarkus and Apache Ignite at Riviera Dev, Devoxx BE, and Devoxx MA in 2023. In 2024, he laid the foundations of what has since become langchain4j-cdi. In 2026, together with Antoine Sabot-Durand, he built Vidocq (https://vidocq.dev) — the first fully European MicroProfile 7.1 runtime.","blueskyUsername":"yblazart.bsky.social","company":"SCIAM","countryName":"France","firstName":"yann","fullName":"yann Blazart","id":18083,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTte0rDSnHTsgef7nTSqwxu70yJKQehjTgrVJ3oSkC7n6hlw=s96-c","lastName":"Blazart","linkedInUsername":"yblazart","mastodonUsername":"","twitterHandle":"yblazart"},{"anonymizedBio":null,"bio":"Sébastien Blanc is a Passion-Driven-Developer with one primary goal : Make the Developers Happy. He likes to share his passion by giving talks that are pragmatic, fun and focused on live coding.","blueskyUsername":"sebi2706.bsky.social","company":"SCIAM","countryName":"France","firstName":"Sebastien","fullName":"Sebastien Blanc","id":21521,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GghdlsQf-LeH_np8EgyP0zMkqdtdA3aeHvPYsYOag=s96-c","lastName":"Blanc","linkedInUsername":"sébastien-blanc-08a73b1","mastodonUsername":"","twitterHandle":"sebi2706"}],"summary":"A candid retrospective on building a Jakarta EE Core Profile and MicroProfile runtime from scratch. It covers architectural tradeoffs, JPMS-driven design changes, CDI event lessons, spec interpretation, dependency minimization, governance decisions, and what they would do differently with JDK 25 and hindsight.","timeSlots":[],"title":"Can AI Build a Jakarta EE Runtime? The Vidocq Experiment","totalFavourites":8,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Most agent talks stop when the agent can call a tool. This session tackles the harder and more practical question: how do you build a custom, production-minded agent for your organization’s workflows and guardrails and make it a first-class participant inside the developer’s IDE?
In this live-coded session, we’ll build a coding agent in Java / Kotlin, connect it directly to an IntelliJ-based IDE through the Agent Client Protocol (ACP), and extend it into an orchestrator that can collaborate with external coding agents such as Codex, Claude, and Gemini.
You’ll learn how to:
- Build your own agent on the JVM, including its reasoning loop, tools, state, and guardrails. Although we’ll create a coding agent, the same techniques apply to chatbots, banking assistants, support agents, and other domain-specific systems.
- Connect a custom agent to an IDE through ACP protocol, without creating a dedicated integration for every agent.
- Integrate and orchestrate multiple external coding agents through ACP in a coordinated pipeline, assigning each task to the agent best suited to perform it.
- Take advantage of the IDE as more than a chat window by giving agents access to the project context, code understanding, developer-defined skills, and more!
By the end of the session, you’ll have a practical blueprint for building an agent that understands your development environment, follows your rules, and can coordinate other agents to complete real software-engineering tasks.","fieldGroup":null,"fieldGroupValues":[],"id":24233,"keywords":[{"name":"guardrails"},{"name":"orchestration"},{"name":"agent"},{"name":"ide"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Technical Product Manager for the ACP Protocol and Agent Ecosystem at JetBrains.
Vadim is the creator and former Technical Lead of the Koog framework at JetBrains. Over the past 9 years at JetBrains, he has contributed to a wide range of projects, including the IntelliJ Kotlin plugin, Kotlin Libraries, the Ktor framework, Qodana Cloud Backend, and Multiplatform Tooling. He also led the technological direction of AI agent development across multiple products as part of the AI Agents Platform.
Today, Vadim is responsible for the ACP Protocol and Open Agent Ecosystem in JetBrains IDEs, driving the vision of making JetBrains IDEs the open gateway for integrating any external AI agents while augmenting them with JetBrains IDEs' deep code understanding and development capabilities.","blueskyUsername":"","company":"JetBrains","countryName":"Hungary","firstName":"Vadim","fullName":"Vadim Briliantov","id":10158,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-08d32376-4f7e-4483-800c-cd8ccf513b2f.jpg","lastName":"Briliantov","linkedInUsername":"vadim-briliantov","mastodonUsername":"","twitterHandle":"VadimBriliantov"}],"summary":"This session shows how to build a production-ready coding agent in Java/Kotlin, connect it to IntelliJ via ACP, and orchestrate external agents like Codex, Claude, and Gemini. It covers reasoning loops, tools, guardrails, IDE context, and coordinated multi-agent workflows for real software engineering tasks.","timeSlots":[],"title":"Build the Builder: A Java Coding Agent in Your IDE with ACP","totalFavourites":null,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"AI coding agents are changing how we develop software. Existing evidence suggests that, up to a point, they significantly reduce the importance of the linguistic features of programming languages, and are able to write code equally well in most if not all mainstream languages. But a programming language is much more than syntax. An AI agent can't make Python code as fast as Java code, and it can't make Java code run as well as JavaScript in the browser. In a world where syntax is less important, the runtime gains even more prominence, and Java's runtime is a superb target for code written by, or with the assistance of, AI.
Java's runtime - the JVM and the standard library - has been Java's core strength since Java's inception. Java programs get faster and easier to operate just by upgrading to a newer JDK, thanks to continually improving JIT optimizations, better GCs, better observability, and a better standard library. This is an especially dynamic time for the Java runtime, integrating value classes from Project Valhalla, AOT code compilation from Project Leyden, making compact object headers the default from Project Lilliput, and removing the need to manually set the heap size so that the JVM could automatically optimise the CPU/RAM tradeoff. Each of these involved tremendous change under the hood, but little or no change to the programming model.","fieldGroup":null,"fieldGroupValues":[],"id":49251,"keywords":[{"name":"java"},{"name":"runtime"},{"name":"jvm"},{"name":"aicoding"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#e89c1a","description":null,"dinerInvite":true,"duration":20,"freeTickets":2,"id":955,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Keynote","pause":false,"scheduleInSmallRoom":false,"slug":"key20"},"speakers":[{"anonymizedBio":null,"bio":"Ron is an architect in the Java Platform Group at Oracle","blueskyUsername":null,"company":"Oracle","countryName":"United Kingdom","firstName":"Ron","fullName":"Ron Pressler","id":49102,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-be6279f3-d001-4451-baa8-03a4ca9af02d.jpg","lastName":"Pressler","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null}],"summary":"AI coding agents reduce the importance of language syntax, making runtime performance and operability more critical. Java’s JVM and standard library are a strong fit: newer JDKs continually improve speed, observability, garbage collection, and usability without changing the programming model, with major innovations underway in Valhalla, Leyden, and Lilliput.","timeSlots":[],"title":"The Year of the Runtime","totalFavourites":11,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"The most agentified engineering teams of my clients are all converging towards the same workflow, aiming to optimize the code review step while preventing human cognitive debt and skill erosion. This talk summarizes their practices, teaching actionable steps and milestones to gradually improve the quality of AI-generated code up to a point in which you can safely start decreasing the code review ‘intensity’. Dark Factories are not built ground up - instead, lights are responsibly dimmed, then turned off one by one, area by area, after observing impeccable code produced consistently.
","fieldGroup":null,"fieldGroupValues":[],"id":31452,"keywords":[{"name":"cognitive debt"},{"name":"ai-generated code"},{"name":"code review"},{"name":"automation"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"With 2 decades of software engineering experience, Victor spent the last 10 years training thousands of developers and architects from 150+ companies. Since early 2026, he’s been working hard to help his clients adopt Agentic Engineering, coaching AI Champions, developers, testers, product owners, ops, and managers with practical, actionable advice to kick-start their agentic reconversion. Every year, Victor gives energetic, intense conference talks in which he distills his training sessions - so be sure not to miss these ideas, heavily brainstormed with millennia of total experience. More about him on https://victorrentea.ro, and https://agentic.how.","blueskyUsername":null,"company":"Freelance","countryName":"Romania","firstName":"Victor","fullName":"Victor Rentea","id":30506,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-51aee41f-0079-4820-98bb-56e70a2e89d3.jpg","lastName":"Rentea","linkedInUsername":"victor-rentea-trainer","mastodonUsername":null,"twitterHandle":"victorrentea"}],"summary":"This talk shares how agentified engineering teams optimize code review for AI-generated code while avoiding cognitive debt and skill erosion. It outlines practical steps and milestones to progressively improve code quality, then safely reduce review intensity as confidence grows, gradually “dimming the lights” area by area.","timeSlots":[],"title":"Responsibly Dimming the Lights in the Software Factory","totalFavourites":31,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"When a nerdy dad and his musician son join forces, some nice things can happen. MelodyMatrix started as a small MIDI experiment and grew into a full project. We presented it in 2024 with a live piano performance at Devoxx and J-Fall to prove it. A year later, AI entered the workflow.
Vik is back at the piano. Frank is back with his laptop. And we bring AI-assisted coding to the stage to demonstrate a workflow that actually worked.
You'll see how AI turned a photo of a hand-drawn chord diagram into a working JavaFX dynamic view, generated a drum set visualization from a description, and how a disciplined Architect/Coder workflow hunted down frame-by-frame rendering differences in Lottie animations. The key insight: design before prompting, keep sessions scoped, and let the failing test tell you what to fix next.
We also used this workflow to improve websites, achieve better SEO results, generate UI views, and burn too many API tokens. You'll see the bill.
No framework benchmarks. No enterprise use cases. Just two people, one piano, a few open-source projects, and an honest look at what systematic AI-assisted coding feels like from the inside.","fieldGroup":null,"fieldGroupValues":[],"id":13752,"keywords":[{"name":"ai-assisted"},{"name":"seo"},{"name":"midi"},{"name":"javafx"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Frank Delporte is a Java Champion working at azul.com, blogger on webtechie.be and foojay.io, author of "Java Programming for Raspberry Pi" (webtechie.be/books), creator of lottie4j.com and melodymatrix.rocks, and contributor to pi4j.com. Frank blogs about his experiments with Java(FX), sometimes combined with electronic components, on the Raspberry Pi.","blueskyUsername":"frankdelporte.be","company":"Azul","countryName":"Belgium","firstName":"Frank","fullName":"Frank Delporte","id":13552,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-fb7af307-de39-456f-a5d2-879c0f97a2f9.jpg","lastName":"Delporte","linkedInUsername":"frankdelporte","mastodonUsername":"https://foojay.social/@frankdelporte","twitterHandle":""},{"anonymizedBio":null,"bio":"16-year full-stack musician with a broad interest in coding with AI tools.","blueskyUsername":null,"company":"The World","countryName":"Belgium","firstName":"Vik","fullName":"Vik Delporte","id":13553,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-3b6109d7-2292-4e6c-8be8-bcfde5babd11.jpg","lastName":"Delporte","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null}],"summary":"MelodyMatrix returns with a live piano demo of AI-assisted coding. The talk shows how AI turned sketches and descriptions into JavaFX and UI visuals, fixed Lottie rendering issues via disciplined testing, and improved websites and SEO. It offers an honest, practical look at systematic AI workflows.","timeSlots":[],"title":"Looking at Music, Again. What a Dad, a Pianist Son, and AI Built Together","totalFavourites":5,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Getting a model to run on a Mac is easy. Getting it to run fast — with production-grade throughput, stable memory behavior, and accuracy that survives optimization — is where the real engineering lives. This session goes deep into the inference stack behind the MLX ecosystem (mlx-vlm, mlx-audio, nativ — 5M+ downloads).
We'll work through the full pipeline with measured numbers at every stage:
Unified memory in practice — why the architecture that makes 512GB of "VRAM" possible also changes how you batch, allocate, and evict; and where memory bandwidth, not compute, becomes the wall across M-series generations.
Continuous batching without a scheduler's luxury — handling concurrent requests when the same RAM serves the OS, the app, and the model.
KV-cache compression and quantization — which techniques hold accuracy on real long-context workloads versus which only look good in benchmarks, with results from MATH500 and LongBench evaluations.
Speculative decoding across chip generations — when draft models pay off and when the overhead eats the win.
Porting war stories — bugs from porting 1000+ models, and the testing methodology that now catches them.
You'll leave able to reason about the accuracy/latency/memory triangle on consumer hardware — and knowing which server-side habits to unlearn.
For engineers building or deploying on-device inference. Familiarity with transformer inference assumed.","fieldGroup":null,"fieldGroupValues":[],"id":31454,"keywords":[{"name":"memory"},{"name":"quantization"},{"name":"latency"},{"name":"inference"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Prince Canuma is Founder and CEO of Neywa Labs, building the inference stack for Apple Silicon. He is the creator and primary maintainer of the MLX open-source ecosystem — mlx-vlm, mlx-audio, and mlx-embeddings — with over 4M combined downloads, foundational to on-device AI on Apple hardware. He previously led developer experience at Neptune AI and worked at Arcee on post-training and inference optimization. His work focuses on model optimization and making AI accessible on local hardware.","blueskyUsername":null,"company":"Neywa Labs","countryName":"Mozambique","firstName":"Prince","fullName":"Prince Canuma","id":16387,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-05bb053f-e043-41b6-af64-c0fcd522a9b7.jpg","lastName":"Canuma","linkedInUsername":"prince-canuma","mastodonUsername":null,"twitterHandle":"Prince_Canuma"}],"summary":"A deep dive into fast, production-grade on-device inference on Macs using the MLX ecosystem. Covers unified memory, continuous batching, KV-cache compression, speculative decoding, and model porting pitfalls, with real benchmark results and engineering tradeoffs for balancing accuracy, latency, and memory on consumer hardware.","timeSlots":[],"title":"Run Frontier Open Models on consumer hardware","totalFavourites":3,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"The world of performance analysis is littered with flawed claims, cognitive biases, dangerous intuitions, and beguiling fallacies. Both Holly and Franz have been guilty of all of the above. Repeatedly. Especially Holly. But this is a no-judgement zone. Some measurement anti-patterns are subtle, and some are downright counter-intuitive. In this talk, Franz and Holly will explain why measuring performance is important, and talk through some of the ways it can go wrong. That would be depressing if that was all there was, so they’ll also introduce a toolbox of questions and principles that you can use to improve the performance of your own applications.
These include:
- How to set up a test system
- Recommended load generators
- The USE method
","fieldGroup":null,"fieldGroupValues":[],"id":14463,"keywords":[{"name":"performance"},{"name":"biases"},{"name":"measurement"},{"name":"load testing"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"I've been working for many years in the computer field. In the last 15 years I've cultivated a strong passion in Java development and the under the hood details of OpenJDK, C and (x86) Assembly.
Big fan of DDD (Domain Driven Design) world, I've developed several Event-Sourcing high performance solutions in the medical and IoT field.
I'm an active member of various online communities on performance (https://groups.google.com/forum/#!forum/mechanical-sympathy), Senior Principal (Software) Performance Engineer and Performance Lead for Red Hat/IBM on Quarkus, Red Hat Top Inventor (2019).
I've collaborated to different projects related high-performance computing both as committer and contributors eg Quarkus, Vert-x, Netty committer, JCTools author, PMC of ActiveMQ Apache Artemis (Messaging Broker), HdrHistogram, JGroups-raft, ...","blueskyUsername":"","company":"IBM","countryName":"Italy","firstName":"Francesco","fullName":"Francesco Nigro","id":5826,"imageUrl":"https://lh3.googleusercontent.com/a/AItbvmlsKs3_saDtZgJ0buTR5LvusOsvlGmdbC7cEeks=s96-c","lastName":"Nigro","linkedInUsername":"","mastodonUsername":"","twitterHandle":"forked_franz"},{"anonymizedBio":null,"bio":"Holly Cummins is a Senior Technical Staff Member on the IBM Quarkus team and a Java Champion. Over her career, Holly has been a full-stack javascript developer, a build architect, a client-facing consultant, a JVM performance engineer, and an innovation leader. Holly has led projects to understand climate risks, count fish, help a blind athlete run ultra-marathons in the desert solo, and invent stories (although not at all the same time). She gets worked up about sustainability, technical empathy, extreme programming, the importance of proper testing, and automating all the things. You can find her at https://hollycummins.com, or follow her on socials at @holly_cummins.
Holly Cummins is a Senior Principal Software Engineer on the Red Hat Quarkus team and a Java Champion. Over her career, Holly has been a full-stack javascript developer, a build architect, a client-facing consultant, a JVM performance engineer, and an innovation leader. Holly has led projects to understand climate risks, count fish, help a blind athlete run ultra-marathons in the desert solo, and invent stories (although not at all the same time). She gets worked up about sustainability, technical empathy, extreme programming, the importance of proper testing, and automating all the things. You can find her at http://hollycummins.com, or follow her on socials at @holly_cummins. ","blueskyUsername":"hollycummins.com","company":"IBM","countryName":"United Kingdom","firstName":"Holly","fullName":"Holly Cummins","id":14657,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-77051efe-f3d4-4e19-b52a-fcf146ad6d77.jpg","lastName":"Cummins","linkedInUsername":"holly-k-cummins","mastodonUsername":"https://hachyderm.io/@holly_cummins","twitterHandle":"holly_cummins"}],"summary":"This talk explores common performance measurement mistakes, cognitive biases, and misleading intuitions. Franz and Holly explain why performance testing matters, how it can go wrong, and offer practical guidance for improving application performance, including test system setup, load generators, and the USE method.","timeSlots":[],"title":"When benchmarks go bad - what we learned from measuring performance wrong","totalFavourites":8,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"This talk explores:
* how classical RSA encryption works
* how quantum algorithms (i.e. Shor’s algorithm) break RSA
* brief overview of where we are regarding quantum hardware
* why PQC still matters right now, to prevent “harvest now, decrypt later” attacks
* what lattice-based PQC looks like and how far along we already are in adopting it
The target audience is everyone generally interested in future technologies. Especially everyone who might have heard about quantum computing and Shor's algorithm breaking encryption, but don't know what that means for the future of cryptography and where we're at today.
The key take aways are a (very basic) general understanding how quantum computing works, to understand how it has the potential to break RSA, as well as quantum-safe encryption methods that are already being put in place, how they work and what the future might look like.","fieldGroup":null,"fieldGroupValues":[],"id":24224,"keywords":[{"name":"quantum"},{"name":"lattice"},{"name":"rsa"},{"name":"pqc"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Senior Developer from Germany with over 15 years of experience building web apps and leading frontend teams, currently living and working down under in Australia. Pragmatic perfectionist, always looking for opportunities to learn and grow.","blueskyUsername":"julianburr.de","company":"Sonar AI","countryName":"Australia","firstName":"Julian","fullName":"Julian Burr","id":24164,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocLOSngLPQFgOpPGXfZqQbR2eBQyiBQsKODaA7Kz896TbM9pjw=s96-c","lastName":"Burr","linkedInUsername":"julianburr","mastodonUsername":"","twitterHandle":""}],"summary":"This talk explains RSA encryption, how Shor’s algorithm could break it, the current state of quantum hardware, and why post-quantum cryptography matters now. It also introduces lattice-based quantum-safe encryption and current adoption progress, helping audiences understand future cryptography and “harvest now, decrypt later” risks.","timeSlots":[],"title":"Post-Quantum Today: How Your Passwords Will Be Safe in a Quantum Future","totalFavourites":9,"track":{"id":1255,"name":"Security, Trust & Compliance","description":"Guardrails, prompt injection defence, model governance, and supply-chain security.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-surveillance-camera.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"Core reflection has been a cornerstone of Java since Java SE 1.1. It provides a unified model for inspecting classes, invoking methods and accessing fields. But reflection is also inefficient. Every reflective call must check access, handle argument conversions like boxing and, finally, invoke the target.
Method handles, introduced in Java 7, address this problem with an execution model that closely mirrors the JVM's. A method handle is one resolved executable operation: a typed capability whose access rights and type are established up front. In Java 9, var handles extended this model to fields and array elements.
In this talk, we’ll show how this trick—moving expensive, one-off setup work out of the hot path—has paid dividends across the Java platform. Method and var handles now sit at the core of many modern runtime features, making string concatenation more efficient, revolutionizing native interop with the Foreign Function & Memory API, and enabling serialization to be redesigned on safer foundations.
This is both an advanced tour of an overlooked part of the Java runtime and a story of resilient platform design: how the right hammer keeps finding new nails.","fieldGroup":null,"fieldGroupValues":[],"id":21666,"keywords":[{"name":"method handles"},{"name":"java runtime"},{"name":"reflection"},{"name":"var handles"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Maurizio Cimadamore is a compiler architect at Oracle. He is currently working on the language and runtime aspects of many foundational projects that will change the shape of the Java platform, such as Project Valhalla and Project Panama.
In the previous years, he has contributed to the design and implementation of several features of the Java programming language, such as diamond inference, lambda expressions and local variable type-inference. Maurizio holds a PhD in language design and implementation.","blueskyUsername":null,"company":"Oracle","countryName":"Italy","firstName":"Maurizio","fullName":"Maurizio Cimadamore","id":21513,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-66579526-c2c5-46b5-87ed-6bb6f34121af.jpg","lastName":"Cimadamore","linkedInUsername":null,"mastodonUsername":null,"twitterHandle":null},{"anonymizedBio":null,"bio":"Viktor is a Software Architect in the Java Platform Group at Oracle.
For more than two decades, he has focused on enabling developers to solve problems with maximal productivity and maintainability.
Specializing in developer ergonomics in the realm of concurrent, parallel, and distributed programming; he has contributed to the standard libraries of several major programming languages, been involved in more than a dozen Open Source projects, and has spoken at numerous conferences and universities.","blueskyUsername":"","company":"Oracle","countryName":"Sweden","firstName":"Viktor","fullName":"Viktor Klang","id":23049,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTtdkY-Lks0gsRcvl05pqpW8-GOpvuVndojO99ZxUsSy0rgo=s96-c","lastName":"Klang","linkedInUsername":"","mastodonUsername":"","twitterHandle":""}],"summary":"The talk explains how Java reflection is inefficient, and how method handles and var handles create reusable, typed capabilities that better match JVM linkage. It also shows how this design supports newer APIs like Foreign Function & Memory and serialization, illustrating resilient platform design.","timeSlots":[],"title":"Teaching an Old Handle New Tricks: From Core Reflection to FFM","totalFavourites":5,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Are you passionate about Java and curious to meet the people shaping its community here in Belgium? Whether you’re a first-time Devoxx attendee, a seasoned Java veteran, or just someone who enjoys geeking out over JVM minutiae, this is the place to be.
Join us for an informal and interactive session. We’ll share what the BeJUG is all about: how we connect Java enthusiasts across the country, the kinds of events we organize (from meetups to hands-on sessions), and how you can get involved as an attendee, a speaker, or even a contributor.
Come say hi, share your ideas, ask questions, or just hang out and network with fellow Java lovers. No slides, no strict agenda, just a room full of like-minded people who care about the Java ecosystem and its vibrant Belgian community. Everyone is welcome, and we’d love to hear what you would like to see from your local Java User Group!","fieldGroup":null,"fieldGroupValues":[],"id":13753,"keywords":[{"name":"java"},{"name":"bejug"},{"name":"community"},{"name":"networking"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":60,"freeTickets":1,"id":954,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"BOF","pause":false,"scheduleInSmallRoom":true,"slug":"bof60"},"speakers":[{"anonymizedBio":null,"bio":"Frank Delporte is a Java Champion working at azul.com, blogger on webtechie.be and foojay.io, author of "Java Programming for Raspberry Pi" (webtechie.be/books), creator of lottie4j.com and melodymatrix.rocks, and contributor to pi4j.com. Frank blogs about his experiments with Java(FX), sometimes combined with electronic components, on the Raspberry Pi.","blueskyUsername":"frankdelporte.be","company":"Azul","countryName":"Belgium","firstName":"Frank","fullName":"Frank Delporte","id":13552,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-fb7af307-de39-456f-a5d2-879c0f97a2f9.jpg","lastName":"Delporte","linkedInUsername":"frankdelporte","mastodonUsername":"https://foojay.social/@frankdelporte","twitterHandle":""},{"anonymizedBio":null,"bio":"Developer Relations Engineer for Timefold, Java Champion and co-organizer of the Belgian Java User Group. Tom has a decade worth of experience delivering systems and loves to share not only knowledge but also passion for our craft.
You can read more at his blog (http://www.tomcools.be) or follow him on Bluesky (@tomcools.be).","blueskyUsername":"tomcools.be","company":"Timefold","countryName":"Belgium","firstName":"Tom","fullName":"Tom Cools","id":13555,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocLiUjyjM49h8fxM3bZDY_EBsphO-BhswQ2kD7N4rUU3lgNYf2Hs=s96-c","lastName":"Cools","linkedInUsername":"tom-cools-17547548","mastodonUsername":"","twitterHandle":""}],"summary":"An informal, interactive session introducing BeJUG, Belgium’s Java User Group. Learn how it connects Java enthusiasts, what events it runs, and how to get involved as attendee, speaker, or contributor. No slides or agenda—just networking, ideas, and community discussion for all Java lovers.","timeSlots":[],"title":"Hi, Belgian Java Community, we want to meet you!","totalFavourites":5,"track":{"id":1253,"name":"People, Practices & Career","description":"Leading teams, engineering culture, and career resilience.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-deepfake-ai.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"Project Amber has introduced a series of Java language features that together
enable a new programming style: data oriented programming. At its core is
compositional pattern matching, which allows values to be deconstructed
naturally and safely.
Where does Project Amber go from here? This talk explores the next wave of
language features under development. These experiments extend pattern matching
in several directions, including user-defined deconstruction for ordinary
interfaces and classes, nested constant patterns for more compact and expressive
matching, an exhaustive pattern-matching statement, safer primitive value
conversions within pattern matching, and a compact notation for deconstructing
and constructing record instances — a process we call reconstruction.
Along the way, I’ll present some of the experimental designs we’re exploring,
discussing both the opportunities they create and the language design challenges
they raise.
","fieldGroup":null,"fieldGroupValues":[],"id":21655,"keywords":[{"name":"reconstruction"},{"name":"deconstruction"},{"name":"java"},{"name":"pattern-matching"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"I work in the Java Platform Group at Oracle designing the Java programming language and virtual machine. I am also a member of the Programming Languages Research Group at Oracle Labs. I am based in Cambridge, UK.
My mission is to keep Java as a leading language and platform for developers around the world. Given my background as a researcher, I am especially keen to find, adapt and transfer ideas and techniques from the research community and get them into the hands of Java programmers. I spend much of my time working on Java’s industry-leading specifications.
I joined Oracle in March 2014. Previously I was a senior researcher at Microsoft Research Cambridge from March 2004 until February 2014. Before that I was a University Lecturer at the University of Cambridge Computer Laboratory and a Fellow, College Lecturer and Director of Studies at St John’s College, Cambridge. I studied for my PhD at the University of Cambridge, and for my BSc at Imperial College, London.","blueskyUsername":"","company":"Oracle","countryName":"United Kingdom","firstName":"Gavin","fullName":"Gavin Bierman","id":21503,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-81f196b1-a89f-4d7a-85d3-5e7e4e171a24.jpg","lastName":"Bierman","linkedInUsername":"gavin-bierman-a0173075","mastodonUsername":"","twitterHandle":"GavinBierman"}],"summary":"Project Amber is advancing Java’s data-oriented programming with new pattern-matching features: user-defined deconstruction, nested constant patterns, exhaustive matching, safer primitive conversions, and compact record reconstruction. The talk explores experimental designs, their benefits, and the language design challenges they introduce.","timeSlots":[],"title":"What’s happening with Project Amber?","totalFavourites":11,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"It feels like with every new Java release, there are more and more Garbage Collectors (GCs) coming with more and more options. We hear cryptic phrases like: "Oh, awesome, ZGC became generational while Shenandoah isn't" or "Did you heard how Netflix reduced their tail latencies with ZGC?"
Naturally, it leaves us with a ton of questions:
- What on earth are they talking about?
- What exactly is ZGC?
- If this ZGC is so magical, why are there even other GCs left in Java anyway?
- Why do we only ever talk about different GCs for Java, but never for Go or JavaScript? Can the Java folks make up their minds and pick one?
- And what about native? Do we even need a GC there?
Whether you are a Java developer or just plain curious, everyone is welcome! We will trace the journey from the creation of Java all the way to the present day to understand why we need a GC and why it isn't a solved problem. We will also take a tour of the various GCs, their evolutions, their strengths, and their weaknesses. (Spoiler alert: ZGC is not the one GC to rule them all).
By the end of this talk, you will know exactly when to use which GC so you can find the perfect fit for every project, every application, and every environment.","fieldGroup":null,"fieldGroupValues":[],"id":9008,"keywords":[{"name":"garbage collection"},{"name":"zgc"},{"name":"java"},{"name":"shenandoah"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"A Java developer with over 20 years of experience, Antoine has been a key part of Axway since 2012.
He dedicated 13 years to their Analytics software, evolving from developer to software architect while maintaining a hands-on development role. His final five years on the project were focused on performance optimization, cloud migration, and security.
Antoine is passionate about open source and reports bugs to community-driven projects. He is always eager to migrate to the latest JDK and introduce new features to his team.
Today, he is leveraging his expertise to build an AI Gateway solution, tackling complex code, and mentoring others.","blueskyUsername":"adessaigne.bsky.social","company":"Axway","countryName":"France","firstName":"Antoine","fullName":"Antoine DESSAIGNE","id":8679,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-763af354-0c0f-4884-9525-88a9034759dd.jpg","lastName":"DESSAIGNE","linkedInUsername":"antoine-dessaigne-3618aa37","mastodonUsername":"","twitterHandle":"adessaigne"}],"summary":"This talk explains why Java has multiple garbage collectors, what ZGC, Shenandoah, and others do, and their strengths and weaknesses. It traces GC history, compares alternatives, and shows when to use each one for the right project, application, or environment.","timeSlots":[],"title":"G1, ZGC, Shenandoah, ... with all these GCs in Java, which one do I choose?","totalFavourites":8,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Structured Concurrency is the next major feature from Project Loom. Following the introduction of Virtual Threads in JDK 21, which dramatically reduced the effort of writing, maintaining, and observing high-throughput concurrent applications, structured concurrency raises the level of abstraction for concurrent programming. It treats groups of related tasks as a single unit of work, simplifying cancellation, error handling, and observability while making concurrent code easier to understand, maintain, and debug.
The feature has evolved through several preview releases, shaped by extensive real-world experience and community feedback. Despite its small surface area, the API enables a wide range of structured concurrency patterns.
In this session, we will explain the motivation and design of structured concurrency, show how to use it effectively, and discuss the lessons learned as the feature evolved. We'll also look beneath the surface at the new lower-level API behind the latest design, offering an early glimpse of work that is expected to form the basis of a future lower-level structured concurrency API.
","fieldGroup":null,"fieldGroupValues":[],"id":29351,"keywords":[{"name":"projectloom"},{"name":"structuredconcurrency"},{"name":"concurrency"},{"name":"virtualthreads"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Alan Bateman is an architect in the Java Platform Group at Oracle and a long-time contributor to OpenJDK. He leads the OpenJDK Core Libraries Group and has helped shape the evolution of the Java platform through the design and implementation of many JDK features, including New I/O (NIO), JVMTI and serviceability, the Java Module System, and, more recently, Project Loom.","blueskyUsername":"","company":"Oracle","countryName":"Ireland","firstName":"Alan","fullName":"Alan Bateman","id":28801,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-777f0d6e-3ce7-4670-9962-7f3e5f309f46.jpg","lastName":"Bateman","linkedInUsername":"","mastodonUsername":"","twitterHandle":""}],"summary":"Structured Concurrency in Project Loom builds on virtual threads to simplify concurrent programming by treating related tasks as one unit. It improves cancellation, error handling, and observability, and makes code easier to understand and debug. The session covers its design, usage, evolution, and future lower-level API work.","timeSlots":[],"title":"Structured Concurrency Looms","totalFavourites":5,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"The LLM scanners like Augustus and Garak are established tools for security evaluation of LLM-based applications, as they are scanning by running through a fixed sample of adversarial attack data, and evaluating a response. However, LLMs are non-deterministic by design, meaning that each input can generate a different response. It is not only non-determinism, but also the vast linguistic attack surface and the wide variety of threats that demonstrate that traditional Unit testing methods are insufficient. This talk proposes a Java tool for security and safety testing for LLM-based applications using probabilistic testing methods, and security contracts. In contrast to the traditional scanners, Tiberius fits naturally into a standard Java test suite, and integrates with the probabilistic testing tool PUnit to support multi-trial scanning and statistical assertions. Different adversarial probes can be integrated and tested directly in Java code, e.g., application specific adversarial attacks, like prompt injections and jailbreaks. A scan (multi-) run can execute 200+ attack probes against a deployed model and serializes the results into versioned json test-fixtures, which can be used in downstream regression tests. The tool furthermore includes a fingerprinting capability, and detects systemic bias, as even system prompts and guardrails can be biased. The audience will leave knowing Java-native LLM security and safety testing techniques.","fieldGroup":null,"fieldGroupValues":[],"id":9024,"keywords":[{"name":"java"},{"name":"security"},{"name":"llm"},{"name":"testing"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Iryna Dohndorf is a software engineer at Karakun, where she works on customer projects using the Java technology stack. With strong technical expertise, genuine curiosity, and a deep passion for computer science, she develops technologies that tackle the challenges of modern companies and today’s society.
An enthusiastic interdisciplinary researcher and developer, Iryna has conducted research in modeling and optimization at TU Dortmund and has worked on software for autonomous driving, government agencies, and EU projects supporting the advancement of organic farming.
Beyond her project work, Iryna is deeply engaged in the international developer community. She is an active member of the BaselOne Program Committee and the Devoxx UK Program Committee,and a frequent speaker at international conferences. Through her talks, writing, and community
involvement, she aims to inspire developers, encourage critical thinking about emerging technologies,
and contribute to the sustainable growth of the global Java ecosystem.
Outside of work, she enjoys spending time with family and friends, and is passionate about swimming, skiing, and—most recently—cycling.
You can learn more about her software development work and research at: https://iryna-dohndorf.netlify.app & https://github.com/tiberius-security/tiberius","blueskyUsername":"","company":"Karakun","countryName":"Germany","firstName":"Iryna","fullName":"Iryna Dohndorf","id":8682,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-ee16aab0-6755-4c6b-8b99-3dd750f954c3.jpg","lastName":"Dohndorf","linkedInUsername":"iryna-dohndorf","mastodonUsername":"","twitterHandle":""}],"summary":"This talk introduces Tiberius, a Java-native tool for LLM security and safety testing. It uses probabilistic multi-trial scans, security contracts, and statistical assertions to test adversarial attacks like prompt injections and jailbreaks, detect bias, and generate versioned JSON fixtures for regression testing.","timeSlots":[],"title":"Beyond assertEquals: Probabilistic Security and Bias Testing for Java LLM Applications","totalFavourites":7,"track":{"id":1255,"name":"Security, Trust & Compliance","description":"Guardrails, prompt injection defence, model governance, and supply-chain security.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-surveillance-camera.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Whether you want to reduce your carbon footprint or cut back on machine resource usage, energy-efficient builds matter more than ever. CI systems often waste energy and time because of inefficient build pipelines. In this lunch session I will demonstrate a few tricks that will give you greener and faster Maven builds. We’ll look at practical examples using GitLab CI and GitHub Actions, so you can apply the techniques right away in your own projects.","fieldGroup":null,"fieldGroupValues":[],"id":24244,"keywords":[{"name":"maven"},{"name":"energy"},{"name":"githubactions"},{"name":"ci"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Jan is an experienced solution architect with over 15 years of experience as Kotlin/Java developer. He loves to combine architecture and coding as this is the sweet spot to invent clever, solutions that have an impact in society.
With a MSc. in Geophysics he has a natural interest in geospatial technology, climate change and maps. He has mainly leveraged these for the public interest in assignments for Dutch government agencies.
In his spare time he practices photography with a focus on street and documentary photography. Living in a university town, he is engaged in a project that supports students and young professionals in the area of well-being and meaning.","blueskyUsername":"janboonen.bsky.social","company":"the/experts.","countryName":"Netherlands","firstName":"Jan","fullName":"Jan Boonen","id":24175,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-83370e03-af8c-4887-818a-857c2813597a.jpg","lastName":"Boonen","linkedInUsername":"janboonen","mastodonUsername":"","twitterHandle":""}],"summary":"This session shows practical ways to make Maven builds greener and faster by improving CI pipelines. Using GitLab CI and GitHub Actions examples, it demonstrates techniques to reduce energy use, save machine resources, and cut build time.","timeSlots":[],"title":"Green Maven builds: sustainable CI in practice","totalFavourites":8,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"TL;DR: DuckDB is a complete analytical SQL engine that runs in-process. No server, no dependencies, one binary, MIT licensed. This session is one duck and a lot of live demos.
While the big data warehouses were busy adding pricing tiers, a small open source project from a research group in Amsterdam quietly built one of the fastest analytical engines around. No license fees, no vendor lock-in, no sales call required.
What makes DuckDB remarkable is how far the same engine scales. It starts as a CLI tool on your laptop, poking at CSV, JSON and Excel files with a friendly SQL dialect and larger-than-memory processing that spills to disk instead of falling over. The same binary embeds into your Java application via JDBC and runs analytics in-process, with no network hop in sight. Point it at S3 and it queries files in place; add DuckLake and it becomes a full lakehouse on open formats. And with the new Quack protocol, ducks are learning to talk to each other, turning DuckDB into a client-server database with multiple writers.
Expect live demos along that entire scale, from laptop to lakehouse.
You will leave with a good sense of what DuckDB can do for you, where its limits are, and why it keeps showing up in so many stacks.
Who is this for? Developers, architects and data-curious engineers who occasionally need answers from data but would rather not operate a warehouse to get them. SQL basics are enough; no data engineering background required.","fieldGroup":null,"fieldGroupValues":[],"id":17558,"keywords":[{"name":"sql"},{"name":"lakehouse"},{"name":"duckdb"},{"name":"analytics"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Oliver Geisser is a freelance Cloud & AI Solution Architect with 20+ years in software development, architecture, and technical leadership. A former Enterprise Architect at an international corporation, he designed and delivered a company-wide cloud-first strategy. Today he helps teams modernize applications, adopt cloud platforms, and bring agentic AI into engineering workflows.
See more: https://www.geisser-consulting.de/","blueskyUsername":"ogeisser.bsky.social","company":"Geisser Consulting","countryName":"Germany","firstName":"Oliver","fullName":"Oliver Geisser","id":16391,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-49e100ac-5a9a-4e96-9a22-85b2a52b50d4.jpg","lastName":"Geisser","linkedInUsername":"oliver-geisser-29ba6879","mastodonUsername":"","twitterHandle":"ogeisser"}],"summary":"DuckDB is a fast, open-source in-process analytical SQL engine with no server or dependencies. This session demos its scale from laptop analytics on CSV/JSON/Parquet to embedded Java apps, S3 queries, and lakehouse use with DuckLake and Quack, aimed at developers and engineers needing simple data analysis.","timeSlots":[],"title":"From Laptop to Lakehouse: Analytical SQL with DuckDB","totalFavourites":4,"track":{"id":1254,"name":"Cloud Native, Data & Infrastructure","description":"Kubernetes, platform engineering, vector DBs, model serving, and GPU orchestration.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-server-settings.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Agentic AI is something everyone wants to use or sell. We wanted an overview of the work and active agents and therefore we created an orchestration tool. Using the Agent, Skills and Spec files from default locations so its interchangeable with other tools. And it really started to change the way we worked on software projects.
As a Java software Engineer it could be seen as a automation of tasks, as an Scrum Master/Product Owner a way to coordinate and for the business analyst a way to store functional requirements. We've used Embabel Agent as the fundamental layer. On top of that we've created a visual representation so we can have a website open in our office which shows which agent is active, what it's doing and what's the next task.
For Software Development teams this will be the biggest change, developers are more reviewing and delivering feedback to the AI while business analysts had a field day of delivering the requirements for the spec files which were reviewed by an AI agent before they could be assigned to an AI-agent. The product owner is in control of the queue where sprints are no longer weeks but days.
I want to share how we created this orchestrator and explain on which impact each technical step had on the way-of-work in our team. It wasn't just about creating an application anymore, everything had impact on how we work. ","fieldGroup":null,"fieldGroupValues":[],"id":23531,"keywords":[{"name":"agentic"},{"name":"requirements"},{"name":"orchestration"},{"name":"automation"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"CTO and Software Architect with over 10 years of hands-on java development experience. He is passionate about Software Development, Artificial Intelligence (AI), Robotics, IoT, Cybersecurity, Open Source Software and loves tech gadgets. Ricky is a speaker on conferences like Code&Comedy, DevSecCon, JFall, Devoxx UK and has a background as a lecturer on an University of Applied sciences. Besides all that Ricky is a Fitness and Martial Arts instructor and enthusiast. Current focus is forming and testing new AI best practices. ","blueskyUsername":"","company":"MetaFactory","countryName":"Netherlands","firstName":"Ricky","fullName":"Ricky van Rijn","id":21529,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-5361a8d1-ea08-4d32-8575-5dbaae4cb186.jpg","lastName":"van Rijn","linkedInUsername":"rickyvanrijn","mastodonUsername":"","twitterHandle":"RickyvanRijn"}],"summary":"We built an agent orchestration tool using Embabel Agent, Skills, and Spec files to coordinate AI-driven software work. It provides a visual dashboard for active agents, tasks, and next steps, reshaping team roles: developers review AI output, analysts define requirements, and product owners manage faster sprint queues.","timeSlots":[],"title":"We created our own AI orchestrator and it became so much more!","totalFavourites":13,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Distributed systems have a dirty secret. Your message broker swears the domain event is published — others are already listening, already reacting to this new order. But then, in a split second, your database rolls back. A constraint violation. Something breaks. And now the entire system knows about an order that never actually existed.
This talk covers the patterns that prevent that chaos: the Dual Write problem, Outbox and Inbox patterns for consistency between database and event stream, idempotent consumers, and Sagas — Orchestration and Choreography.
You'll leave knowing which pattern solves which problem, when to use each, and what you're really trading off with eventual consistency","fieldGroup":null,"fieldGroupValues":[],"id":17791,"keywords":[{"name":"distributedsystems"},{"name":"inbox"},{"name":"sagas"},{"name":"outbox"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Hey there, I'm Emanuel - a Software Craftsman from Romania.
I value simplicity and intentionality over blindly applying patterns and abstractions.
I care deeply about the design and testability of the software I create.
I also value communication and collaboration and enjoy sharing my knowledge through open-source contributions and technical articles.
","blueskyUsername":"","company":"","countryName":"Romania","firstName":"Emanuel","fullName":"Emanuel Trandafir","id":18092,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocJIeD6DlrYMLqQ25N-MsDC0Oz2dracw1hlS6BqbbRBwEXJINS13=s96-c","lastName":"Trandafir","linkedInUsername":"emanueltrandafir","mastodonUsername":"","twitterHandle":""}],"summary":"This talk explains how to avoid inconsistencies in distributed systems caused by dual writes and failed transactions. It covers Outbox/Inbox patterns, idempotent consumers, and Saga orchestration/choreography, showing when to use each and the trade-offs involved with eventual consistency.","timeSlots":[],"title":"Consistency Patterns in Event-Driven Architecture","totalFavourites":12,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"We bring a real legacy Scala application built on Play and Akka on stage. Anonymized from production code that we have worked on and migrate it live to Java 25 twice. We use Claude as the coding agent, but the method works with any agent. Willem Jan steers it toward Quarkus and Radek explores Spring Boot. We take turns instead of racing side by side, so the room can follow one decision at a time. We stop at each checkpoint where we score both the framework and the agent's output on a shared scorecard.
The session has four parts:
- Diagnose and plan (together): a tour of the legacy app, why a big-bang rewrite is off the table, the strangler fig approach, and characterization tests written with the agent before we touch anything.
- Translate the domain (together): case classes, Option, Either, and for-comprehensions become records, sealed interfaces, pattern matching, and Optional. The agent proposes, while we review and correct the suggestions live.
- Two apps (alternating): guiding both frameworks through a series of migration stages. At each checkpoint, we unpack what went right, where the agent needed human intervention, and update our scorecard across performance, developer experience, and testability.
- Verdict (together): the final scorecard, when we would pick which framework, where the agent actually saved time and where it needed a human override, and open Q&A
","fieldGroup":null,"fieldGroupValues":[],"id":21670,"keywords":[{"name":"migration"},{"name":"java"},{"name":"quarkus"},{"name":"spring"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"Radek has a background in Software Engineering, ML, and AI from his undergraduate and postgraduate studies at TU Delft, where he recently completed his Master's Thesis in Explainable AI (XAI) for large language models (LLMs). He sees great potential in integrating and understanding how ML/AI can be woven into modern software, a passion he's brought to stages across Europe, speaking at conferences like Devoxx, Voxxed Days, JNation, and LunaConf on topics ranging from secure AI agents with Quarkus LangChain4j to ML development for software engineers.
But Radek doesn't stop at code. He spends his mornings and evenings making people sweat as a group fitness instructor, and his talk Debugging Your Health: Fitness for Desk Warriors has resonated with audiences across multiple conferences. With a passion for tech and fitness, he loves helping desk warriors optimize both their code and their health. When he's not debugging software or coaching group workouts, you can find him experimenting with new workout routines, geeking out over biomechanics, or trying to convince his coworkers to do squats during stand-up meetings.","blueskyUsername":"","company":"Lunatech","countryName":"Netherlands","firstName":"Radek","fullName":"Radek Kargul","id":3944,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocJzV9KZ_Fhk3y62Yn8Kg8JHArGLGXsjsYInr82h8yopG5O5QJ0=s96-c","lastName":"Kargul","linkedInUsername":"radek-kargul","mastodonUsername":"","twitterHandle":""},{"anonymizedBio":null,"bio":"Willem Jan is a Principal Software Engineer at Lunatech who loves building cloud software and contributing to open source. He's all about writing quality code and finding ways to automate anything. Willem Jan has been working with Quarkus since the beginning and enjoys making sure his applications are both simple and secure. On weekends, you won't find him behind a keyboard - he's either out enjoying nature or volunteering with his local scout group.","blueskyUsername":"wjglerum.nl","company":"Lunatech","countryName":"Netherlands","firstName":"Willem","fullName":"Willem Jan Glerum","id":19752,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-fdb3dada-6952-4509-96e0-c39c4c5571e1.jpg","lastName":"Jan Glerum","linkedInUsername":"wjglerum","mastodonUsername":"","twitterHandle":""}],"summary":"A live modernization of a legacy Play/Akka app on Java 25: first protected by characterization tests, then translated from Scala to modern Java. The same slice is rebuilt twice—once in Quarkus, once in Spring Boot—while comparing startup, memory, DI, testability, and trade-offs.","timeSlots":[],"title":"Legacy JVM modernization with agentic AI: Quarkus vs Spring Boot","totalFavourites":5,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"After several years of Java development, I had left the ecosystem behind. Not out of spite, but out of pragmatism: Python was getting the job done. Beefy CI/CD pipelines, easier to write; daily tools that bent to fit any need; in short, the simple joy of writing a Hello World without filling out three XML files first. The rise of AI cemented Python as the obvious choice. I had no reason left to reopen a Java IDE. None.
Except that a handful of friends kept the thread alive, with that community, with that ecosystem. A conference here, an article there, and little by little, certain projects started catching my eye. Step by step, I discovered what you could now build CLIs in Java with picocli, with a type system that doesn't lie on top. Then came Quarkus: blazing-fast startup, native compilation, no more excuses about JVM heft. The final piece of the trinity was TamboUI: elegant TUIs and a pleasant DSL, offering a modern, refreshing take on terminal applications, in Java.
In this talk, let me tell you how Java found its way back into my toolbox and my geeky heart: why I had left, why I thought I'd never come back, and how the community kept my mind open. No sermon, no holy war, just a language and technology lover reconsidering things, and who thinks you might want to take a second look too.
","fieldGroup":null,"fieldGroupValues":[],"id":10802,"keywords":[{"name":"java"},{"name":"quarkus"},{"name":"tamboui"},{"name":"picocli"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"A passion born from reading, the world of computing and electronics caught me from my ZX Spectrum and has never let go.
A developer by trade, from video games to master data management, I've also dabbled in the worlds of QA and production deployment.
I'm now a Developer Advocate at Dataiku, juggling between tech, community, and knowledge sharing.
Since a passion always deserves to be passed on, I founded Mixteen to try to help children grasp the digital world.
This desire to share has also led me to take part in organizing several MiXiT conferences and then to join the board of the AFPy.
In parallel, I also try to share through blog posts and videos on my YouTube channel.
In short:
https://calm.show/talks/
In detail:
https://www.twitch.tv/titimoby
https://mixteen.org/
https://tontoncodeur.fr/
https://www.youtube.com/thierrychantier
https://www.afpy.org
","blueskyUsername":"titimoby.eu","company":"Dataiku","countryName":"France","firstName":"Thierry","fullName":"Thierry Chantier","id":10180,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocJIqLuDzHLmlJCLNs_i3ZQmzlennTw_KQgt3Dz4qdDy_JF2_w=s96-c","lastName":"Chantier","linkedInUsername":"thierrychantier","mastodonUsername":"https://mamot.fr/deck/@TitiMoby","twitterHandle":""}],"summary":"A developer explains leaving Java for Python, then rediscovering Java through picocli, Quarkus, and TamboUI. The talk explores Java’s modern CLI, fast native apps, and terminal UIs, showing how community and new tooling made Java relevant again.","timeSlots":[],"title":"A Deserter Returns: What the Java community helped me rediscover","totalFavourites":null,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Understanding how to monitor and optimize Java applications is fundamental to building efficient software.
This hands-on lab will introduce you to capturing data to understand your application behavior using JDK Flight Recorder (JFR), managing and analyzing JFR recordings, and creating custom JFR events. We'll also cover how JFR can integrate with Prometheus to correlate application and infrastructure behavior, all aimed at enhancing observability.
After this lab, you'll have the confidence to use JFR to monitor, diagnose, and fine-tune your applications.","fieldGroup":null,"fieldGroupValues":[],"id":15360,"keywords":[{"name":"observability"},{"name":"java"},{"name":"jfr"},{"name":"prometheus"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#00FF00","description":null,"dinerInvite":true,"duration":120,"freeTickets":2,"id":984,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":2,"name":"Hands-on Lab","pause":false,"scheduleInSmallRoom":true,"slug":"lab120"},"speakers":[{"anonymizedBio":null,"bio":"Ana is a Java Champion Alumni and Senior Developer Advocate for the Java Platform Group at Oracle, with an extensive background in Java and cloud development. More recently, her work has expanded to explore how machine learning and AI can be applied within the Java ecosystem, helping Java developers adopt emerging capabilities while continuing to benefit from the stability, tooling, and performance of the Java Platform. She actively supports the growth of technical communities through knowledge sharing and enjoys curating content for conferences as a program committee member. ","blueskyUsername":"ammbra1508.bsky.social","company":"Oracle","countryName":"Netherlands","firstName":"Ana-Maria","fullName":"Ana-Maria Mihalceanu","id":5830,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14Gj1OS6OpAzuwBKwoUuCi9gxvNPvdmVA7nbW8lg3mg=s96-c","lastName":"Mihalceanu","linkedInUsername":"","mastodonUsername":"https://mastodon.social/@ammbra1508","twitterHandle":"ammbra1508"},{"anonymizedBio":null,"bio":"José is a Java Developer Advocate at Oracle. He holds a PhD in applied mathematics and computer science and is a Java Champion alumnus. He has been an assistant professor at the University Sorbonne Paris Nord for 25 years. He is an active member of the Paris Java User Group, and a disorganizer of the JChateau unconference. He contributes Java learning content on the dev.java website. He publishes "Java Coding Tips", a biweekly series of shorts, and the "JEP Café", a video series, both on the Java YouTube channel.","blueskyUsername":"josepaumard.bsky.social","company":"Oracle","countryName":"France","firstName":"José","fullName":"José Paumard","id":15304,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GhsVIUuTR9Onoo-D8UMW5QId2Yw1V_EUvqdWdNQ0RQ=s96-c","lastName":"Paumard","linkedInUsername":"jos%C3%A9-paumard-2458ba5","mastodonUsername":"","twitterHandle":""}],"summary":"Hands-on lab on Java observability using JDK Flight Recorder (JFR). Learn to capture, manage, and analyze recordings, create custom events, and integrate JFR with Prometheus to correlate application and infrastructure behavior. Participants will gain practical skills to monitor, diagnose, and optimize Java applications.","timeSlots":[],"title":"Analyze and Optimize Your Applications with JFR","totalFavourites":11,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Generative AI can write code in seconds. So does it still make sense to teach programming to children?
We at Devoxx4Kids think it does.
In our workshops, we use LEGO SPIKE robots to teach problem-solving, software design and computational thinking. There is just one challenge: LEGO SPIKE does not officially support Java.
Rather than accepting that limitation, we explored alternative approaches to bring Java to the platform and make LEGO robots accessible to our next generation Java developers.
In this session, we'll share our journey, demonstrate how Java can be used to control LEGO SPIKE robots, discuss the technical challenges we encountered, and show real-world examples from educational settings.
Attendees will learn:
- Why programming fundamentals remain relevant in the AI era
- How LEGO SPIKE can be integrated into a Java ecosystem
- Technical approaches for controlling SPIKE robots with Java
- Lessons learned while teaching children with physical computing platforms
Expect live demos, moving robots, and perhaps a few LEGO bricks on stage.","fieldGroup":null,"fieldGroupValues":[],"id":10045,"keywords":[{"name":"java"},{"name":"lego spike"},{"name":"programming education"},{"name":"generative ai"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Eddy is the chairman of Devoxx4Kids Netherlands and an Engineering Manager at Ahold Delhaize Netherlands. He has a focus on AI, Data, FinOps and Java.
He loves to share his knowledge and help in his free time, with the Devoxx4Kids Foundation, kids to learn how to code in a fun way.","blueskyUsername":"eddyvos.bsky.social","company":"Devoxx4Kids Netherlands","countryName":"Netherlands","firstName":"Eddy","fullName":"Eddy Vos","id":10179,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-9b520e91-967e-42a4-9061-ff04e1ff6682.jpg","lastName":"Vos","linkedInUsername":"eddyvos","mastodonUsername":"","twitterHandle":"eddyvos_EVCS"}],"summary":"This session shows why teaching programming still matters in the AI era, using LEGO SPIKE robots in Java-based workshops for children. It covers technical methods to control SPIKE with Java, challenges encountered, educational lessons learned, and includes live robot demos.","timeSlots":[],"title":"Java, LEGO & AI: Teaching the Next Generation to Build, Not Just Prompt","totalFavourites":3,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"The Loom project is about making asynchronous programming simpler. The promise is to enable the "simple thread-per-request style to scale with near-optimal hardware utilization". Before virtual threads, this was only achievable with callback-based reactive programming. Loom first brought virtual thread in JDK 21 that makes the blocking of threads almost free. Along with it, it published the Scoped Values API in JDK 25, as a replacement of the Thread Local API. The last element is the Structured Concurrency API, which is still a preview feature in JDK 26 and 27. This hands-on lab guides you through writing a virtual thread-based web application that utilizes structured concurrency to parallelize requests, and implements different strategies to process them. Then it shows you how you can use scoped values to pass sensitive information from one module of your application to another one, without relying on method parameters. It also uses the principles of Data Oriented Programming to organize your application in a simple way. You will use StructuredTaskScopes, the implementations of the Joiner interface that the API gives you, and you will create your own implementations as well. In the end, you will see the patterns of code that the Loom project gives you to achieve the same throughput as a classical reactive application, and you will then be able to compare both approaches to choose which one you prefer.","fieldGroup":null,"fieldGroupValues":[],"id":22920,"keywords":[{"name":"structuredconcurrency"},{"name":"scopedvalues"},{"name":"virtualthreads"},{"name":"loom"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#00FF00","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":968,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":2,"name":"Hands-on Lab","pause":false,"scheduleInSmallRoom":true,"slug":"lab18"},"speakers":[{"anonymizedBio":null,"bio":"Ana is a Java Champion Alumni and Senior Developer Advocate for the Java Platform Group at Oracle, with an extensive background in Java and cloud development. More recently, her work has expanded to explore how machine learning and AI can be applied within the Java ecosystem, helping Java developers adopt emerging capabilities while continuing to benefit from the stability, tooling, and performance of the Java Platform. She actively supports the growth of technical communities through knowledge sharing and enjoys curating content for conferences as a program committee member. ","blueskyUsername":"ammbra1508.bsky.social","company":"Oracle","countryName":"Netherlands","firstName":"Ana-Maria","fullName":"Ana-Maria Mihalceanu","id":5830,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14Gj1OS6OpAzuwBKwoUuCi9gxvNPvdmVA7nbW8lg3mg=s96-c","lastName":"Mihalceanu","linkedInUsername":"","mastodonUsername":"https://mastodon.social/@ammbra1508","twitterHandle":"ammbra1508"},{"anonymizedBio":null,"bio":"José is a Java Developer Advocate at Oracle. He holds a PhD in applied mathematics and computer science and is a Java Champion alumnus. He has been an assistant professor at the University Sorbonne Paris Nord for 25 years. He is an active member of the Paris Java User Group, and a disorganizer of the JChateau unconference. He contributes Java learning content on the dev.java website. He publishes "Java Coding Tips", a biweekly series of shorts, and the "JEP Café", a video series, both on the Java YouTube channel.","blueskyUsername":"josepaumard.bsky.social","company":"Oracle","countryName":"France","firstName":"José","fullName":"José Paumard","id":15304,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GhsVIUuTR9Onoo-D8UMW5QId2Yw1V_EUvqdWdNQ0RQ=s96-c","lastName":"Paumard","linkedInUsername":"jos%C3%A9-paumard-2458ba5","mastodonUsername":"","twitterHandle":""}],"summary":"A hands-on lab on Project Loom’s virtual threads, structured concurrency, and scoped values. It teaches building a scalable web app, parallelizing requests, passing sensitive data safely, and comparing Loom’s thread-per-request approach with reactive programming for similar throughput.","timeSlots":[],"title":"Using Structured Concurrency and Scoped Values to Create a Virtual Threads Based Asynchronous Application","totalFavourites":8,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Every architecture decision carries a sustainability trade-off. A greener data center may sit farther from users, increasing latency. Going serverless eliminates idle compute waste but introduces cold starts and the complexity of maintaining multiple functions.
This session explores the tension between sustainability and the other "-ilities" (maintainability, portability, availability, performance) and presents practical techniques to reduce energy use, carbon footprint, and operating costs.
You will leave the session with a concrete catalog of trade-offs you can apply to your next project.","fieldGroup":null,"fieldGroupValues":[],"id":8169,"keywords":[{"name":"sustainability"},{"name":"trade-offs"},{"name":"latency"},{"name":"serverless"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Ioannis Kolaxis is a Director and a Senior Technology Architect at Accenture. He focuses on Green Software innovations, optimizing applications to have lower costs, reduced energy consumption and fewer carbon emissions. He is an inventor, speaker, and author of the Green Software book.
He was a kid in primary school when he first saw the green screen with the blinking cursor of an Amstrad CPC6128. This computer opened a new world to him, a world of continuous exploration and creation through programming.
Ioannis invented a novel way to work remotely and holds four patents for software solutions.
He frequently speaks at conferences such as GOTO Amsterdam 2022, Devoxx Ukraine 2021 and Oracle Code One 2019.
Visit his website to learn more: https://kolaxis.dev","blueskyUsername":"","company":"Accenture","countryName":"Greece","firstName":"Ioannis","fullName":"Ioannis Kolaxis","id":7644,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocJPsPavZKJoz3GJkmJ0SRkTbEPd96EJotlWK3x9Fl082Dw=s96-c","lastName":"Kolaxis","linkedInUsername":"ioannis-kolaxis","mastodonUsername":"","twitterHandle":""}],"summary":"The session examines sustainability trade-offs in architecture decisions, balancing energy use and carbon footprint against maintainability, portability, availability, and performance. It offers practical techniques to reduce emissions and costs, and provides a catalog of trade-offs to apply in future projects.","timeSlots":[],"title":"How to Architect Green Software","totalFavourites":13,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Debugging a consistently failing unit test is easy; you set a breakpoint and fix it. But how do you debug a "random" failure in a massive microservices landscape or a distributed monolith?
During an interview, I was once told, "You are more of a debugger than a coder." They were right: I thrive on the bugs that make others give up: the nasty, intermittent issues hidden under the carpet.
But you cannot solve these issues with 'System.out.println()' or a simple debugger attached to localhost.
Debugging at scale requires a shift in mindset and tooling. It requires moving from the paradigm of "5 Whys" and 'logger.info' to a structured observational approach based on observability data.
Based on production outages, I'll share how I uncovered the problems, the lessons I learned and how you can combine a debugging mindset with observability tooling like OpenTelemetry and the Grafana toolset to gain deeper insights.","fieldGroup":null,"fieldGroupValues":[],"id":21152,"keywords":[{"name":"observability"},{"name":"debugging"},{"name":"grafana"},{"name":"opentelemetry"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Observability enthusiast and Grafana Champion, always looking for ways to improve software with better observability and what new insights can be gained. Delivering reliable software is always my priority. Combining existing tools often results in new ways to get an even better view. I work for OpenValue as a software & observability engineer and SRE.","blueskyUsername":"ceesbos.nl","company":"OpenValue","countryName":"Netherlands","firstName":"Cees","fullName":"Cees Bos","id":16354,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-14d6453f-7d33-4f7a-86e4-3568aac637c9.jpg","lastName":"Bos","linkedInUsername":"ceesbos","mastodonUsername":"https://foojay.social/@cbos","twitterHandle":""}],"summary":"This talk explores debugging elusive, intermittent failures in large microservice and distributed systems. It argues for a shift from traditional logging and local debugging to structured observability, using tools like OpenTelemetry and Grafana to uncover root causes and gain deeper system insight.","timeSlots":[],"title":"Beyond system.out.println: Applying a Debugging Mindset at Scale","totalFavourites":10,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"The most powerful compute fleet ever assembled wasn't built by a cloud provider. It's the half a billion-plus Apple Silicon devices already in laps, pockets, and offices — each with a GPU, a neural engine, and unified memory that would have qualified as a workstation five years ago. Collectively they represent more silicon than any datacenter on Earth. And for AI, they sit almost entirely idle.
This keynote makes the case that the next era of AI won't be defined by bigger clusters, but by inference moving to where the data already lives. I'll tell that story through building the MLX open-source ecosystem (mlx-vlm, mlx-audio, nativ — 5M+ downloads, used by Google DeepMind, Hugging Face, and LM Studio): what it actually takes to make consumer hardware run frontier models, why the server playbook fails on-device, and the hard-won engineering — quantization that holds accuracy, KV-cache compression, speculative decoding across chip generations — that closes the gap.
You'll leave with a new mental model for where AI workloads belong: which belong in the cloud, which belong on-device, and why the answer is about to flip for more of them than you think, for reasons of privacy, latency, cost, and sovereignty.
For every engineer and leader deciding where their AI features will live.","fieldGroup":null,"fieldGroupValues":[],"id":31455,"keywords":[{"name":"applesilicon"},{"name":"quantization"},{"name":"on-device"},{"name":"inference"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#e89c1a","description":null,"dinerInvite":true,"duration":20,"freeTickets":2,"id":955,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Keynote","pause":false,"scheduleInSmallRoom":false,"slug":"key20"},"speakers":[{"anonymizedBio":null,"bio":"Prince Canuma is Founder and CEO of Neywa Labs, building the inference stack for Apple Silicon. He is the creator and primary maintainer of the MLX open-source ecosystem — mlx-vlm, mlx-audio, and mlx-embeddings — with over 4M combined downloads, foundational to on-device AI on Apple hardware. He previously led developer experience at Neptune AI and worked at Arcee on post-training and inference optimization. His work focuses on model optimization and making AI accessible on local hardware.","blueskyUsername":null,"company":"Neywa Labs","countryName":"Mozambique","firstName":"Prince","fullName":"Prince Canuma","id":16387,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-05bb053f-e043-41b6-af64-c0fcd522a9b7.jpg","lastName":"Canuma","linkedInUsername":"prince-canuma","mastodonUsername":null,"twitterHandle":"Prince_Canuma"}],"summary":"This keynote argues that Apple Silicon devices form a massive, underused AI compute fleet. Drawing on the MLX ecosystem, it shows how on-device inference can handle frontier models through advanced optimization, and why privacy, latency, cost, and sovereignty will shift more AI workloads from cloud to device.","timeSlots":[],"title":"Half of a billion Idle GPUs: The Case for On-Device AI","totalFavourites":9,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Every time you ship a Java service to Kubernetes, you ship a base image, a bundled JDK, and an OS layer. None of which you wrote besides your application code and the selection of dependencies, but all of which you must monitor and regularly patch. The real artifact you care about, your application JAR and its dependencies, are buried inside 250 MB of packaging you never wanted to own. WebAssembly already solved this. With KWasm, the Wasm runtime lives on the node, the developer ships only a .wasm module as an OCI artifact, and a RuntimeClass tells Kubernetes to execute it. There is no Dockerfile, no container image involved at all.
Most Kubernetes users know that "runtimeClassName: kata" or "runtimeClassName: wasm" routes a pod to a different executor. Few know how this works with the containerd Runtime v2 TTRPC protocol, OCI bundle assembly, overlay filesystems, and runc delegation that make all this possible.
In this talk we walk through how the KWasm analogy actually maps cleanly to Java, and what custom containerd shims, node provisioners, and CRDs can actually do: allow you to build your own Kubernetes-based platform (and quite easily with agentic AI). We will explore a new open source project called Brewlet.sh with live coding: its CLI brewlet push target/app.jar, the Kubernetes deployment approach kubectl apply, and see in practice how a Spring Boot service runs on Kubernetes without a single container image in sight.","fieldGroup":null,"fieldGroupValues":[],"id":25048,"keywords":[{"name":"kubernetes"},{"name":"springboot"},{"name":"containerd"},{"name":"webassembly"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Bruno Borges is a Principal Product Manager at GitHub with over 30 years of experience in Software Engineering, Product Management, Developer Relations, and Cloud & AI developer experience. Currently, he leads developer relations and product management programs for AI and GitHub Copilot at Microsoft, driving initiatives like the GitHub Copilot SDK for Java, and AI-assisted modernization workflows for enterprise developers. Bruno also continues to shape Microsoft's Java strategy, having led the public announcement of the Microsoft Build of OpenJDK and new tools like Azure Command Launcher for Java. Prior to his current focus on AI, he spent years building Microsoft Azure into a top destination for Java workloads, and before that served nearly six years at Oracle influencing developer relations strategy. A recognized Java Champion and top Java influencer, Bruno is a frequent conference speaker and open-source contributor with projects like Java Evolved, JairoSVG, JLib Inspector, ghx, and more. Visit github.com/brunoborges","blueskyUsername":"brunoborges.bsky.social","company":"GitHub","countryName":"Canada","firstName":"Bruno","fullName":"Bruno Borges","id":3933,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-32df7003-16c8-41d9-9052-accdde8c4fdf.jpg","lastName":"Borges","linkedInUsername":"brunocborges","mastodonUsername":"","twitterHandle":"brunoborges"}],"summary":"The talk shows how KWasm-style runtime delegation can let Java apps run on Kubernetes without container images. It explains containerd internals, custom shims, node provisioners, and CRDs, then demonstrates Brewlet.sh, a new open-source project that deploys a Spring Boot JAR directly as an OCI artifact.","timeSlots":[],"title":"Java on Kubernetes Without Container Images: The WebAssembly Way","totalFavourites":9,"track":{"id":1254,"name":"Cloud Native, Data & Infrastructure","description":"Kubernetes, platform engineering, vector DBs, model serving, and GPU orchestration.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-server-settings.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Can you create a complete game with just 10 lines of code? That was the quest I set myself when recreating the classic game Snake in as few lines as possible. Let me tell you: I was not ready for the types of challenges I would face. For example, just registering button presses was not as straightforward as I had hoped.
Through live coding, I will show you how I overcame these obstacles and distilled Snake to its absolute essence. What key elements make this game instantly recognizable? By only using tools from the JDK, we will end up with a minimal code base and if not award-winning, at least a playable version of Snake.
Striving for conciseness led me to a deeper understanding of the code. I will show how constraints push us to think outside the box and lead to more creative solutions. Moreover, my approach can be applied to any problem. Sometimes the best way to understand code is by seeing how little of it you actually need!","fieldGroup":null,"fieldGroupValues":[],"id":19748,"keywords":[{"name":"coding"},{"name":"snake"},{"name":"minimalism"},{"name":"constraints"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Guus is a Java/Kotlin engineer at adesso Netherlands. With a passion for puzzles and a background in science, he strives for elegant solutions to complex problems. In his spare time, he likes to over-optimize the simple tasks of today, to hopefully live a lazy life tomorrow.","blueskyUsername":"","company":"adesso","countryName":"Netherlands","firstName":"Guus","fullName":"Guus de Wit","id":19777,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-df4a0ee3-439b-464e-94f4-33a5dce8a655.jpg","lastName":"de Wit","linkedInUsername":"guus-de-wit-0846a5161","mastodonUsername":"","twitterHandle":""}],"summary":"A live-coding talk recreates Snake in 10 lines of JDK-only code, overcoming input and design challenges to distill the game’s essence. The project demonstrates how extreme conciseness fosters creativity, deeper understanding, and transferable problem-solving techniques.","timeSlots":[],"title":"Snake in 10 Lines: Learning More by Coding Less","totalFavourites":8,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Can Java build a real AI agent — one that manages your calendar, reads your email, orders pizza, and remembers who you are across sessions? OpenClaw, the personal AI agent with 350K GitHub stars, proves the concept. We're going to build it twice, in Java, live on stage.
Baruch brings JetBrains Koog, Viktor brings LangChain4j. Same features, same LLM, completely different philosophies. We'll run six competitive rounds of coding, from basic agent setup through memory, tool calling, agentic workflows, guardrails, and observability. Each round surfaces a design disagreement: should memory be an Advisor or a Provider? Are agents composed services or first-class citizens? And when your guardrail framework and the model disagree, who wins?
The frameworks disagree on how AI agents should be built. The audience votes on who's right.","fieldGroup":null,"fieldGroupValues":[],"id":25027,"keywords":[{"name":"java"},{"name":"frameworks"},{"name":"ai"},{"name":"agent"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"Viktor Gamov is a Developer Advocate at Confluent, a company that makes a cloud streaming platform based on Apache Kafka.
Viktor has extensive expertise in building enterprise application architectures using open-source technologies.
He enjoys helping architects and developers design and develop low-latency, scalable, and highly available distributed systems.
He is a professional conference speaker on distributed systems, streaming data, the JVM, DevOps, and other topics, and regularly speaks at events including JavaOne, Devoxx, OSCON, QCon, and others.
He co-authored O’Reilly’s «Enterprise Web Development.»
He blogs at http://gamov.io and co-hosts DevRelRad.io.
Follow Viktor on Twitter @gamussa, where he posts about gym life, food, open-source, and, of course, Kafka and Confluent!","blueskyUsername":"gamussa.dev","company":"Confluent","countryName":"United States of America","firstName":"Viktor","fullName":"Viktor Gamov","id":9605,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTtd9ZkVQ9giwUn-Tq1GRAjZ3ip6tMZ0Jnr2XMv2ECSYvkwJO=s96-c","lastName":"Gamov","linkedInUsername":"vikgamov","mastodonUsername":null,"twitterHandle":"gamussa"},{"anonymizedBio":null,"bio":"Baruch Sadogursky (@jbaruch) did Java before it had generics, DevOps before there was Docker, and DevRel before it had a name. He built DevRel at JFrog from a ten-person company through an IPO, co-authored "Liquid Software" and "DevOps Tools for Java Developers," and is a Java Champion, a Microsoft MVP for AI, and a CNCF Ambassador alum.
Today, as a head of DevRel at Port and a self-certified Level 3 Context Sommelier, Baruch focuses on context engineering. On top of sharing context with AI agents, Baruch also shares knowledge with developers through blog posts, meetups, and conferences such as DevNexus, QCon, Kubecon, and Devoxx, mostly about how to make your AI engineering lifecycle smarter and increase your organization's productivity.","blueskyUsername":"jbaru.ch","company":"Port.io","countryName":"United States of America","firstName":"Baruch","fullName":"Baruch Sadogursky","id":9606,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocKqAphxO-bPI7xrWrgh3Vhn_yJ1H91TCUKCNY_tGTQagjntFox7kA=s96-c","lastName":"Sadogursky","linkedInUsername":"jbaruch","mastodonUsername":"","twitterHandle":"jbaruch"}],"summary":"A live Java showdown builds the same AI agent twice using JetBrains Koog and LangChain4j. Across six coding rounds, the speakers compare memory, tool calling, workflows, guardrails, and observability, exposing different design philosophies. The audience votes on which framework best fits real agent development.","timeSlots":[],"title":"Codepocalypse Now: LangChain4j vs JetBrains Koog","totalFavourites":8,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Curious about the evolution and future direction of LangChain4j? Join our Birds of a Feather (BOF) session for an open and interactive discussion with members of the LangChain4j community. Since its debut at Devoxx Belgium 2023, LangChain4j has rapidly grown into a key framework for integrating large language models into Java applications. Whether you're already using LangChain4j in production or just starting to explore its capabilities, this session is a great opportunity to connect with fellow developers, contributors, and enthusiasts. We'll discuss the current state of the project, real-world adoption stories, ongoing challenges, and exciting new developments on the roadmap. This is your chance to ask questions, share feedback, exchange ideas, and help shape the future of LangChain4j. Come with curiosity and leave with insights, connections, and inspiration!","fieldGroup":null,"fieldGroupValues":[],"id":16435,"keywords":[{"name":"java"},{"name":"llm"},{"name":"langchain4j"},{"name":"framework"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":60,"freeTickets":1,"id":954,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"BOF","pause":false,"scheduleInSmallRoom":true,"slug":"bof60"},"speakers":[{"anonymizedBio":null,"bio":"Dimitris is an open source veteran with 25+ years of experience in IT and he is currently a (Red Hat) Sr.Director of Engineering at IBM in charge of the Cloud Native Runtimes group that includes Quarkus, Vert.x, SpringBoot and Node.js. He also served as the JBoss AS project lead and he has been a JBoss addict and contributor from the early start-up days. He worked previously at Intracom and Motorola in the areas of NMS/OSS, designing reusable frameworks and distributed systems. Dimitris studied computer science at the Technological Educational Institute of Athens and received an M.Sc. by research from University College Dublin, Ireland.
Dimitris is a passionate advocator of Opensource & Red Hat technologies and has presented on a variety of events and conferences around the world. He has been an invited speaker at various Java/Linux community events and user groups, and he has delivered inspirational talks on technology and Open Source in universities and educational institutes. In his spare time he has also helped with the organization of technology enablement events for children, like Scratch Days, Devoxx4Kids and JCrete4Kids.","blueskyUsername":"","company":"IBM","countryName":"Switzerland","firstName":"Dimitris","fullName":"Dimitris Andreadis","id":10167,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-c540ac38-0fdf-40cb-b0fe-8a357abd1f87.jpg","lastName":"Andreadis","linkedInUsername":"dandreadis","mastodonUsername":"","twitterHandle":"dandreadis"}],"summary":"An open BOF session on LangChain4j’s evolution, adoption, challenges, and roadmap. Join community members to discuss real-world use, share feedback, ask questions, and help shape the future of this fast-growing Java framework for integrating large language models.","timeSlots":[],"title":"LangChain4j Community BOF","totalFavourites":4,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Let's face it... containers and Kubernetes are everywhere!
But developers don't necessarily want to become Kubernetes specialists.
This presentation will focus on what developers need to know to get the most of Kubernetes without having to become a DevOps.
We will cover:
- modern local development environment (flox, kind...)
- building container images (multistage build, build without docker...)
- important kubernetes features developers need to know and how to handle them efficiently (requests/limits, graceful shutdown, probes, OCI volumes...)
- how to investigate and debug code on kubernetes (kubectl debug, mirrord...)
- how to secure applications (SecurityContext, CEL, SBOM, RuntimeClass with gVisor/Kata...)
- how to integrate observability with OpenTelemetry
The presentation will contain examples and demos with a strong focus on Best Practices.","fieldGroup":null,"fieldGroupValues":[],"id":17759,"keywords":[{"name":"usage"},{"name":"companies"},{"name":"kubernetes"},{"name":"evolution"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Alain Regnier is a Technical Architect and Entrepreneur passionate about innovation and new technologies.
He is a Kubestronaut and GDE Cloud (Google Developer Expert).
He works as Technical Director of the Kubernetes & DevOps group at SCIAM, a company providing Expertise Consulting Services.
Before that he was CTO for Kubo Labs, a company specialized in Kubernetes on premise and on the Cloud.
He helps companies that want to move to Kubernetes or use it more efficiently following Best Practices. (strategy, cluster installation, applications update, training, audit...)
He has also spent 10 years in Silicon Valley working for startups and large companies, where he co-authored various standards about Web Services and Connected Devices.","blueskyUsername":"","company":"SCIAM","countryName":"France","firstName":"Alain","fullName":"Alain Regnier","id":18056,"imageUrl":"https://lh3.googleusercontent.com/a-/AFdZucoMvtrrwfCGhvKIFfqBfRpLpaU_hbL3lYUoE_Dulw=s96-c","lastName":"Regnier","linkedInUsername":"alainregnier","mastodonUsername":"","twitterHandle":""}],"summary":"Kubernetes for developers: what you need to know, without becoming a Kubernetes expert.","timeSlots":[],"title":"Kubernetes for Developers","totalFavourites":9,"track":{"id":1254,"name":"Cloud Native, Data & Infrastructure","description":"Kubernetes, platform engineering, vector DBs, model serving, and GPU orchestration.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-server-settings.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Everyone is talking about AI-assisted development, but what does that mean for the next generation?
Are we still teaching kids to code? Should we? What skills matter when AI generates code? How do we make technology attractive and accessible to children?
Join organizers, mentors, teachers, parents, and developers for an open discussion on the future of Devoxx4Kids, coding education, robotics, and growing the next generation of makers.","fieldGroup":null,"fieldGroupValues":[],"id":10046,"keywords":[{"name":"coding"},{"name":"robotics"},{"name":"ai-assisted"},{"name":"education"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":60,"freeTickets":1,"id":954,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"BOF","pause":false,"scheduleInSmallRoom":true,"slug":"bof60"},"speakers":[{"anonymizedBio":null,"bio":"Eddy is the chairman of Devoxx4Kids Netherlands and an Engineering Manager at Ahold Delhaize Netherlands. He has a focus on AI, Data, FinOps and Java.
He loves to share his knowledge and help in his free time, with the Devoxx4Kids Foundation, kids to learn how to code in a fun way.","blueskyUsername":"eddyvos.bsky.social","company":"Devoxx4Kids Netherlands","countryName":"Netherlands","firstName":"Eddy","fullName":"Eddy Vos","id":10179,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-9b520e91-967e-42a4-9061-ff04e1ff6682.jpg","lastName":"Vos","linkedInUsername":"eddyvos","mastodonUsername":"","twitterHandle":"eddyvos_EVCS"}],"summary":"An open discussion on how AI-assisted development affects children’s coding education, including whether kids should still learn to code, which skills matter most, and how to make technology engaging and accessible. The session explores the future of Devoxx4Kids, robotics, and nurturing the next generation of makers.","timeSlots":[],"title":"Who will build software after AI?","totalFavourites":5,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Last year I built Backlog.md, an open source tool that captures your intent, splits it into small markdown tasks, and lets you follow your agents' progress on a TUI Kanban board. A year and 300,000 agent-written lines later, my projects had quietly turned into a small dark factory: software getting built with nobody really reviewing it. Task tracking told me what the agents did, but not what my system had become. I am no longer the owner of my code.
Today I am building Groma.md to fix this. Groma keeps a real-time architectural blueprint of your project's domain as the source of truth, and tasks become the means to reach the blueprint's next iteration. Imagine UML without the fear that it goes stale, where you can watch exactly which parts of the system your agents are modifying.
Whether you are adding an order management system to a brownfield codebase or starting a new project from scratch, you design it in Groma and let the agents write the code, while you keep an eye on the overall architecture and stay in control of the meaningful decisions.
In this talk I will take a large open source repository and map it into a Groma blueprint live on stage, to show you how to become the owner of your code again.","fieldGroup":null,"fieldGroupValues":[],"id":25652,"keywords":[{"name":"architecture"},{"name":"blueprint"},{"name":"kanban"},{"name":"agents"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Alex Gavrilescu leads backend & web engineering at Funstage GmbH in Vienna, keeping millions of free‑to‑play gamers happily tapping. He still ships code, tinkers with Raspberry Pi Kubernetes clusters for fun, and is passionate about weaving project‑management smarts with practical AI. Most recently he created Backlog.md, a micro‑tool that turns side‑project chaos into shippable tasks.","blueskyUsername":"","company":"Backlog.md","countryName":"Italy","firstName":"Alex","fullName":"Alex Gavrilescu","id":24199,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-05b8ab40-661f-4685-ba72-9774462ba13b.jpg","lastName":"Gavrilescu","linkedInUsername":"alexandrugavrilescu","mastodonUsername":"","twitterHandle":"mrlesk"}],"summary":"Groma.md is an open-source tool that maintains a live architectural blueprint of your codebase, keeping you in control while agents write code. Unlike task tracking, it shows how the system evolves in real time. The talk will demo mapping a large repository into a Groma blueprint live.","timeSlots":[],"title":"In the loop, not in the dark: staying in control in the era of dark software factories","totalFavourites":8,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"You’ve probably seen AI agents in action—reasoning through complex problems, orchestrating API calls, recovering from mistakes, and maintaining intent across dozens of interactions. But how do they actually work? This workshop answers that question from first principles. Through a series of progressively more capable examples, we’ll evolve a simple request/response interaction into a complete agentic system with an explicit control loop: plan → act → observe → decide. Rather than treating agents as black boxes, we’ll expose the architectural patterns that make them reliable, understandable, and testable.
The examples are implemented with Spring Boot and Spring AI, but this is not a framework tutorial. The focus is on the fundamental patterns that underpin agentic systems, regardless of language, framework, or AI model. Using a dinner planning assistant as our running example, we’ll progressively introduce tool invocation, planning, memory, multi-step execution, evaluation, error handling, and termination conditions, showing how each capability emerges naturally as the agent evolves.
A central theme of the workshop is that agents are systems embedded in systems. Building robust agents is less about prompt engineering and more about software architecture. By the end of the session, you’ll have a clear mental model of the agentic loop and a practical set of architectural patterns you can apply to build production-ready agentic systems, regardless of the technologies you use.","fieldGroup":null,"fieldGroupValues":[],"id":24247,"keywords":[{"name":"memory"},{"name":"architecture"},{"name":"planning"},{"name":"agents"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":180,"freeTickets":2,"id":959,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Deep Dive","pause":false,"scheduleInSmallRoom":false,"slug":"deep180"},"speakers":[{"anonymizedBio":null,"bio":"Adib Saikali is a Distinguished Engineer at Tanzu, focused on helping large enterprises design and build AI-enabled platforms and applications using Spring. His work centers on AI agent architectures, agentic control loops, API design for autonomous systems, and enterprise governance for AI in production. Adib is the author of Software Security for Developers and brings over 25 years of experience across startups and global enterprises in roles spanning developer, architect, and CTO.","blueskyUsername":null,"company":"Broadcom","countryName":"Canada","firstName":"Adib","fullName":"Adib Saikali","id":24178,"imageUrl":"https://lh3.googleusercontent.com/a/AAcHTtfY6Rruj1Zj96IlogKnZ3KQRWlq0jO9-eEuvaq7QgiQhw=s96-c","lastName":"Saikali","linkedInUsername":"asaikali","mastodonUsername":null,"twitterHandle":"asaikali"}],"summary":"This workshop explains agentic AI from first principles, evolving a simple request/response flow into a full plan-act-observe-decide loop. Using Spring Boot/Spring AI examples, it covers tools, planning, memory, multi-step execution, evaluation, error handling, and termination, emphasizing robust software architecture over prompt engineering.","timeSlots":[],"title":"Building AI Agents from First Principles","totalFavourites":8,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Building Production-Ready A2A Agents with LangChain4j, Quarkus, Spring Boot & Jakarta EE
Enterprise AI is shifting from isolated chatbots to multi-agent ecosystems where specialized agents negotiate and delegate tasks. The Agent-to-Agent (A2A) Protocol is the emerging standard for this interoperability, defining the wire format for seamless cross-language communication.
This intensive, hands-on workshop equips Java developers to build, connect, and scale production-grade A2A ecosystems using LangChain4j and the official A2A Java SDK.
🛠️ What You Will Learn & Build
The journey begins with the core architecture, where you will decode the A2A specifications and implement an AgentCard for schema discovery alongside an AgentCardExecutor for standardized message delivery.
From there, you will transition into a multi-runtime build, deploying identical agent capability contracts natively across Spring Boot, JA2A, and Quarkus.
To prove true language independence, you will connect a Python-based agent built with the A2A Python SDK to achieve seamless, cross-framework interaction.
As the network grows, you will dive into orchestration by building a dynamic router agent capable of decomposing complex, multi-turn tasks and dispatching them across your diverse mesh.
Finally, you will prepare for Enterprise Day-2 Operations by mastering production challenges like distributed state management, load balancing for stateful LLM conversations, and tracing complex interactions via OpenTelemetry (OTel).","fieldGroup":null,"fieldGroupValues":[],"id":5516,"keywords":[{"name":"microprofile"},{"name":"generativeai"},{"name":"jakartaee"},{"name":"langchain4j"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#00FF00","description":null,"dinerInvite":true,"duration":120,"freeTickets":2,"id":984,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":2,"name":"Hands-on Lab","pause":false,"scheduleInSmallRoom":true,"slug":"lab120"},"speakers":[{"anonymizedBio":null,"bio":"Into the Java World for the last 25 years, I joined Red Hat in 2013 to contribute on the development of the JakartaEE server WildFly and moved to IBM in 2025.
I also contribute to a large number of OSS projects and I'm a member of the Apache NetBeans Dream Team.
I've been one of the founder of the SnowCamp conference and I'm a JUG leader of AlpesJUG.
Recently I've started to look into Generative AI and LangChain4J and I started the wildfly-ai-feature-pack to facilitate developing Generative AI JakartaEE applications with WildFly, contributing to langchain4j-cdi and LangChain4J along the way.
I also contribute to the A2A Java SDK providing a reference implementation of the A2A protocol in Java.","blueskyUsername":"ehsavoie.bsky.social","company":"IBM","countryName":"France","firstName":"Emmanuel","fullName":"Emmanuel Hugonnet","id":5813,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocI_LNcfEMO7A9N-kuKXYVnc6dpm2KSwRws_s0f4sSklrbdqruzA=s96-c","lastName":"Hugonnet","linkedInUsername":"ehsavoie","mastodonUsername":"https://framapiaf.org/@ehsavoie","twitterHandle":"ehsavoie"},{"anonymizedBio":null,"bio":"With over 25 years of experience in enterprise software development and AI engineering, I explore how artificial intelligence is transforming the way we build, learn, and live with technology. I have an old passion for Open Source and more recent one for AI Engineering.","blueskyUsername":"","company":"IBM","countryName":"Italy","firstName":"Stefano","fullName":"Stefano Maestri","id":16360,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocLGEYxm84uctahd-qUUcxCFJlvsJdS1ob8UEiziPUNLeb2CmlFF=s96-c","lastName":"Maestri","linkedInUsername":"meste","mastodonUsername":"","twitterHandle":"maeste"},{"anonymizedBio":null,"bio":"Mario is a senior principal software engineer at IBM working as Drools project lead. Among his interests there are also high performance systems and generative AI, being an active contributor of widely adopted projects like Quarkus and LangChain4j. He is also a Java Champion, the JUG Milano coordinator, a frequent speaker and the co-author of "Modern Java in Action" published by Manning.","blueskyUsername":null,"company":"IBM","countryName":"Italy","firstName":"Mario","fullName":"Mario Fusco","id":3932,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GinuLQCwL67BBgfGsiWRl3mu3Ya0TbDffhYFiXX=s96-c","lastName":"Fusco","linkedInUsername":"mario-fusco-3467213","mastodonUsername":null,"twitterHandle":"mariofusco"}],"summary":"LangChain4j-CDI brings generative AI to Jakarta EE/MicroProfile with a CDI extension for easy AI service injection. It demonstrates building enterprise AI agents using MCP and A2A protocols, plus resilience, observability, and multi-provider configuration. Live coding shows deployment on Quarkus, WildFly, and Helidon in minutes.","timeSlots":[],"title":"Wiring Up the AI Mesh: Building A2A-Compliant Java Agents","totalFavourites":null,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"This interactive session transforms AI security education into an engaging courtroom style debate. We'll present five critical vulnerabilities affecting modern AI development tools, and after each case study, the audience becomes the jury—voting on who's responsible: the careless user, the negligent developer, or the inadequate service provider.
We'll dissect real CVEs including GitHub Camo Leak (CVE-2025-59145), Ollama RCE (CVE-2024-37032), AWS MCP Server Shell Injection (CVE-2025-5277), Anthropic MCP Inspector DNS Rebinding (CVE-2025-49596 ) and Microsoft Copilot's zero-click data exfiltration (CVE-2025-32711). Each case reveals technical root causes through collaborative analysis. The List of CVEs will be usually update based on the new related CVEs as well.
The conclusion challenges the "blame game" itself: these vulnerabilities expose fundamental architectural weaknesses in agentic AI systems where traditional security models fail. We will talk about Mental Model Gap and how this phenomenon has been seen before in the technological leaps before. We'll establish that securing AI tools demands a shared responsibility framework developers must code defensively, providers must architect securely, and users must understand AI-specific risks.","fieldGroup":null,"fieldGroupValues":[],"id":7020,"keywords":[{"name":"shared responsibility"},{"name":"cve"},{"name":"vulnerabilities"},{"name":"ai security"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Makan Sepehrifar is a software architect who is interested in how people think and behave in organizations. He sees software architecture as a craft and believes that every challenge has both a technical side and a social side, and both need attention at the same time. Coming from a strong developer background, he has led many teams across different domains and organizations, always working to connect technology with human needs and team dynamics.
He is passionate about the future of Developer Experience and how Generative AI and Agentic AI are reshaping the way developers work. Makan focuses on how these technologies can improve developer satisfaction and help teams become more adaptive and effective. He is a public speaker on Developer Experience practices and AI in this space, and he has been part of several Generative AI adoption teams in different companies, helping them use AI in a practical and effective way.","blueskyUsername":"makan1869.bsky.social","company":"Productboard","countryName":"Netherlands","firstName":"Makan","fullName":"Makan Sepehrifar","id":5847,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-68014c00-8dd0-4467-9ac6-f205258d440d.jpg","lastName":"Sepehrifar","linkedInUsername":"makan1869","mastodonUsername":"","twitterHandle":""}],"summary":"An interactive courtroom-style session on AI security, where audiences judge responsibility for five real CVEs in AI tools. It analyzes root causes, evolving vulnerabilities, and the “blame game,” arguing that securing agentic AI requires shared responsibility among developers, providers, and users.","timeSlots":[],"title":"Who to blame? The AI, The Programmer, or The Prompt?","totalFavourites":8,"track":{"id":1255,"name":"Security, Trust & Compliance","description":"Guardrails, prompt injection defence, model governance, and supply-chain security.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-surveillance-camera.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"So you got "promoted" to lead a team / project / whatever. What now? Do you have what it takes? It certainly seems that your hierarchy thinks so. But oftentimes things are not so black and white. Being a great plumber doesn't automatically make you a great leader of plumbers, so you need to adapt and adjust your ways of working by dropping old habits and picking up new ones. Then you start getting worried by the fact that you won't be able to have as much impact as before, since you're not coding as much as you used to in the past (and that's what you're good at, right?).
What if leadership can amplify your technical impact instead? In this talk, I’ll share how I evolved from individual contributor to a mix of Solution Architect, Tech Lead and classical Engineering Manager at CERN - keeping my hands in the code while scaling my influence through the team that I work with. I'll be sharing best practices and adjustment techniques for you to use when faced with the difficult transition process into technical leadership. And lastly, we'll be talking about how to stay relevant even if you're not spending countless hours honing your coding skills anymore.
","fieldGroup":null,"fieldGroupValues":[],"id":9025,"keywords":[{"name":"leadership"},{"name":"technical"},{"name":"influence"},{"name":"transition"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Dr. Cristian Schuszter holds a PhD in Systems Engineering with a focus on distributed fault-tolerant software architectures and machine-learning based failure prediction. In his day-to-day, he is somewhere between a Data and Full-stack Software Engineer, with a wide range of tech & job exposure both in academia and industry.
For the past 8 years he's been working with various teams in CERN (the European Organization for Nuclear Research). Currently he's focusing on the architectural directions of the CERN business computing group, with an emphasis on enterprise Java systems and UI.
","blueskyUsername":"cschuszter.bsky.social","company":"CERN","countryName":"Switzerland","firstName":"Cristian","fullName":"Cristian Schuszter","id":8687,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-8a031ed5-f55f-4758-8655-819459e29ce6.jpg","lastName":"Schuszter","linkedInUsername":"cschuszter","mastodonUsername":"","twitterHandle":""}],"summary":"This talk explores the transition from individual contributor to technical leader, showing how leadership can amplify impact rather than reduce it. Drawing on experience at CERN, it shares practical techniques for adapting, staying relevant, and maintaining technical influence while leading teams.","timeSlots":[],"title":"The Quantum Tech Lead: Hands-On and Abstract at the Same Time","totalFavourites":7,"track":{"id":1253,"name":"People, Practices & Career","description":"Leading teams, engineering culture, and career resilience.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-deepfake-ai.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Project Leyden raises a hard question for Quarkus: can we benefit from faster JVM startup and warm-up without undoing the assumptions that make Quarkus fast today? After years of optimizations to reduce startup costs, profiling JVM Quarkus applications still points to familiar hotspots: class loading and reading from JAR files.
Quarkus has already pushed hard in these areas through build-time processing, custom packaging, and class loading strategies. Leyden promises a major shift for JVM applications, but combining it with Quarkus is not just about enabling a new feature. Some of Leyden’s current constraints interact directly with optimizations Quarkus relies on, forcing us to question where responsibilities should sit between the framework, the JVM, and the application packaging model.
This talk shares findings from experimenting with Project Leyden inside Quarkus, and you will learn what to watch for when fine-tuning startup and warm-up behavior, and how to reason about the trade-offs between Quarkus’ existing optimizations and Leyden’s emerging JVM-level capabilities. Expect concrete findings, performance observations, current limitations, experimental prototypes, and practical lessons for developers who want to prepare their Quarkus applications for the next generation of JVM startup optimization.","fieldGroup":null,"fieldGroupValues":[],"id":5510,"keywords":[{"name":"optimization"},{"name":"startup"},{"name":"quarkus"},{"name":"leyden"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Georgios works for IBM as a Senior Principal Software Engineer and is currently one of the most active contributors to Quarkus, where he works in all sorts of areas, including but not limited to LangChain4j, RESTEasy Reactive, Spring compatibility, Kubernetes support, testing, Kotlin, and more.
He is also an enthusiastic promoter of Quarkus who never misses a chance to spread the Quarkus love!","blueskyUsername":null,"company":"IBM","countryName":"Greece","firstName":"Georgios","fullName":"Georgios Andrianakis","id":3945,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-95904645-1695-4adf-b24c-41926050b4b4.jpg","lastName":"Andrianakis","linkedInUsername":"georgios-andrianakis","mastodonUsername":null,"twitterHandle":"geoand86"},{"anonymizedBio":null,"bio":"Ana is a Java Champion Alumni and Senior Developer Advocate for the Java Platform Group at Oracle, with an extensive background in Java and cloud development. More recently, her work has expanded to explore how machine learning and AI can be applied within the Java ecosystem, helping Java developers adopt emerging capabilities while continuing to benefit from the stability, tooling, and performance of the Java Platform. She actively supports the growth of technical communities through knowledge sharing and enjoys curating content for conferences as a program committee member. ","blueskyUsername":"ammbra1508.bsky.social","company":"Oracle","countryName":"Netherlands","firstName":"Ana-Maria","fullName":"Ana-Maria Mihalceanu","id":5830,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14Gj1OS6OpAzuwBKwoUuCi9gxvNPvdmVA7nbW8lg3mg=s96-c","lastName":"Mihalceanu","linkedInUsername":"","mastodonUsername":"https://mastodon.social/@ammbra1508","twitterHandle":"ammbra1508"}],"summary":"This talk shares hands-on Quarkus experiments with Project Leyden, exploring how Leyden’s JVM startup and warm-up improvements interact with Quarkus’ build-time optimizations. It highlights what works, current limitations, performance findings, broken assumptions, and design changes needed to combine both approaches effectively.","timeSlots":[],"title":"Quarkus Meets Leyden at the JVM Performance Edge","totalFavourites":2,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"As AI-assisted coding evolves from simple autocomplete to autonomous agents, developers need new mental models and tools to maintain productivity and trust. In this session, we are going to revisit the major developments in agentic coding from over the past few months, from MCP to skills and beyond; discuss how to improve trust in your agents and take a tour of the new Google Antigravity ecosystem. Every technique discussed will be paired with a live demo using practical examples of day to day development so we can see the coding agent in action. Supporting materials for this session can be found at: https://github.com/danicat/state-of-agentic-coding
","fieldGroup":null,"fieldGroupValues":[],"id":25671,"keywords":[{"name":"ai-assisted"},{"name":"autonomous"},{"name":"softwareengineering"},{"name":"agents"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Daniela Petruzalek is an experienced IT professional with background in software engineering, pre-sales and developer relations, currently a Senior Developer Relations Engineer at Google. Her specialisation is data engineering and back end development and she is a former Google Developer Expert in Go and Google Cloud Platform. She's also a Google Cloud Certified Data Engineer, Oracle Certified Professional and TEDx speaker. In her spare time she contributes to open source, plays video games and pets random cats on the streets.","blueskyUsername":"danicat83.bsky.social","company":"Google Cloud","countryName":"United Kingdom","firstName":"Daniela","fullName":"Daniela Petruzalek","id":24186,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocJALSmaAeQ2KPyjkQq0mv6gykwBqgZXif1g3KZyFjCqQDqQZOpEUw=s96-c","lastName":"Petruzalek","linkedInUsername":"petruzalek","mastodonUsername":"","twitterHandle":"danicat83"}],"summary":"This session reviews recent advances in agentic coding, from MCP to skills, and explores ways to build trust in AI coding agents. It includes live demos of practical development workflows and an overview of the Google Antigravity ecosystem.","timeSlots":[],"title":"State of Agentic Coding","totalFavourites":null,"track":{"id":1701,"name":"Development Practices","description":"All about Getting Stuff Done right. Methodologies, better practices, etc.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-hearts-card.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"AI coding agents are rapidly becoming first-class citizens in enterprise software workflows. Teams are already delegating consequential decisions to autonomous agents: generating database migrations, refactoring service boundaries, modifying schemas, producing runbooks, and calling external tools. But as agent autonomy grows, a critical question emerges — who is responsible when something goes wrong?
This talk explores the discipline of Human-in-the-Loop (HITL) design, and how to build AI-assisted workflows that keep humans meaningfully in control without sacrificing the productivity gains that make agents worth adopting in the first place. It also examines the growing body of EU regulation that makes HITL not just a good engineering practice, but a legal requirement for certain classes of systems.
The session will demonstrate these concepts through a realistic microservices scenario — a Java enterprise platform with AI-assisted delivery — showing how the absence of human oversight leads to cascading failures, and how a well-designed HITL architecture, grounded in both engineering discipline and regulatory awareness, would have prevented them. Attendees will leave with a concrete framework for moving fast with AI agents without losing control of their systems or their compliance posture.","fieldGroup":null,"fieldGroupValues":[],"id":7017,"keywords":[{"name":"hitl"},{"name":"compliance"},{"name":"autonomy"},{"name":"microservices"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"From programming microservices to leading engineering teams: this is how I've spent my last 20 years in technology as a Java/Spring ecosystem specialist. My career has evolved from designing cloud-native architectures and migrating legacy systems to leading engineering teams focused on high-quality results and efficient processes.","blueskyUsername":null,"company":"IG","countryName":"Spain","firstName":"Juan","fullName":"Juan Antonio Breña 🇪🇸🇪🇺","id":5848,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-40c3fd7e-fe96-414c-b6c4-4b52f1bd300a.jpg","lastName":"Antonio Breña 🇪🇸🇪🇺","linkedInUsername":"juanantoniobrenamoral","mastodonUsername":null,"twitterHandle":"juanantoniobm"}],"summary":"The talk explains Human-in-the-Loop design for AI coding agents in enterprise workflows, covering approval gates, audit trails, compliance with the EU AI Act, and failure risks in microservices. It offers practical patterns for adding human checkpoints to CI/CD without sacrificing speed or control.","timeSlots":[],"title":"The Importance of HITL to Avoid Chaos and EU Regulations for AI","totalFavourites":18,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Secure by Design does not fail only because developers lack security awareness. It often fails because teams prescribe the wrong intervention: training when the real blocker is time, tools, motivation, or trust. In this practitioner-focused talk, we introduce SECUR-E, the OWASP Security Culture Project v2.0 framework that uses COM-B to help developers and software teams diagnose what actually blocks secure development behavior.
Through a relatable story, practical examples, and six memorable developer patterns, Skeptic, Enthusiast, Compliant, Unaware, Resistant, and Embedded, the audience will learn why one-size-fits-all security programs miss, bore, or backfire. We will show how a lightweight self-assessment maps Capability, Opportunity, and Motivation into team insights, then translates those insights into better interventions: evidence for Skeptics, support for Enthusiasts, autonomy for Resistant profiles, foundational help for Unaware developers, and amplification for Embedded security-minded builders.
Key takeaways: stop asking “why don’t developers care?”; diagnose whether the blocker is skill, environment, or motivation; choose interventions that fit the pattern; protect psychological safety; and use SECUR-E as a practical playbook for building secure software teams without blame. Audience: software developers, tech leads, security champions, and engineering managers with beginner-to-intermediate knowledge.","fieldGroup":null,"fieldGroupValues":[],"id":12758,"keywords":[{"name":"security culture"},{"name":"com-b"},{"name":"psychological safety"},{"name":"secure development"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Enrique Larios Vargas is a Security and Learning Specialist with over 8 years of experience designing impactful learning and enablement programs across fintech, engineering, and security domains. With a background as a university lecturer in software engineering in Peru, the Netherlands, and Canada, he brings a unique blend of technical insight and behavioral science to his work. Enrique is the lead author of the research paper “DASP: A Framework for Driving the Adoption of Software Security Practices”, which explores how behavioral models like COM-B can drive secure development. He is passionate about helping developers move beyond compliance and build a meaningful, human-centered security culture.","blueskyUsername":"","company":"OWASP","countryName":"Netherlands","firstName":"Enrique","fullName":"Enrique Larios Vargas","id":12611,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-3a74a43d-dc7b-46a1-84da-7840cb0dff8f.jpg","lastName":"Larios Vargas","linkedInUsername":"enrique-larios-vargas","mastodonUsername":"","twitterHandle":""}],"summary":"SECUR-E, an OWASP framework using COM-B, helps teams diagnose why secure development fails—whether due to skills, environment, or motivation—not just awareness. It identifies six developer patterns and guides tailored interventions, improving secure software practices without blame.","timeSlots":[],"title":"Knocking on the Wrong Door: The Six Developers You’ll Meet in Security","totalFavourites":7,"track":{"id":1253,"name":"People, Practices & Career","description":"Leading teams, engineering culture, and career resilience.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-deepfake-ai.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"It arrives in a box, in pieces, and every part of it is open: hardware, firmware, simulator, SDK. An hour with a screwdriver and there is a robot on your kitchen table. What it does not arrive with is a mind.
So I bought one, and I wrote it a brain called AURA. No lab, no team, just a laptop and some evenings.
AURA knows the people in my house and treats each of them differently. What we told it stays separate from what it worked out on its own. Given a task, it opens VS Code, puts Claude Code to work on a real repository, and asks my permission before anything is committed.
It still gets things wrong. Nothing in the design promises otherwise; what it promises is that nothing sensitive happens without me saying yes.
The robot is not the point. The point is how little it now takes to build one, and that quietly changes who gets to build anything at all. I would rather show you that than argue it, so there will be demos, and no promises about all of them behaving.
","fieldGroup":null,"fieldGroupValues":[],"id":50102,"keywords":[{"name":"firmware"},{"name":"autonomy"},{"name":"privacy"},{"name":"robot"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#e89c1a","description":null,"dinerInvite":true,"duration":20,"freeTickets":2,"id":955,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Keynote","pause":false,"scheduleInSmallRoom":false,"slug":"key20"},"speakers":[{"anonymizedBio":null,"bio":"Jan Van Wassenhove is the Technical Director of the System Integrations Business Line for the BeLux region at Sopra Steria, where he drives technology solutions and services, supported by his team of architects. With a strong background in Java development, Jan has evolved into an experienced architect and project manager, with in-depth expertise in Agile methodologies.
Remarkably, he is also the initiator of “SCRUM”, a new programming language he conceived. This creation perfectly reflects his innovative mindset and constant drive to explore new technological territories — and proves, with a wink, that everyone really can become a SCRUM master… in code.
Throughout his career, he has contributed to numerous projects in both the public and private sectors, with a particular focus on banking and public administrations.","blueskyUsername":"","company":"Sopra Steria","countryName":"Belgium","firstName":"Jan","fullName":"Jan Van Wassenhove","id":7632,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-e7b52977-dc36-455d-a814-67a86c6644ab.jpg","lastName":"Van Wassenhove","linkedInUsername":"jan-van-wassenhove-9b49893","mastodonUsername":"","twitterHandle":""}],"summary":"A hobbyist built AURA, a personal “brain” for an open robot platform, using only a laptop. It recognizes household members, keeps learned and told information separate, and seeks permission before committing code. The project demonstrates how accessible robotics and AI now are.","timeSlots":[],"title":"From Brainless to Brilliant — Giving an Open-Source Robot a Mind of Its Own","totalFavourites":null,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"ADVANCED","description":"Java is full of subtleties, pitfalls, and rules we think we know until we have to predict what a piece of code will actually print.
In this interactive session, we will walk through a series of short Java snippets to analyze together.
The goal is to learn, revisit core concepts, challenge assumptions, and better understand how the Java language really behaves.","fieldGroup":null,"fieldGroupValues":[],"id":10043,"keywords":[{"name":"subtleties"},{"name":"java"},{"name":"pitfalls"},{"name":"snippets"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":60,"freeTickets":1,"id":954,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"BOF","pause":false,"scheduleInSmallRoom":true,"slug":"bof60"},"speakers":[{"anonymizedBio":null,"bio":"Passionate about software development since getting my first programmable calculator, I have built a solid experience as a tech lead and architect in designing embedded combat systems.
An advocate of clean code, "developer joy" and well-thought-out architectures, I enjoy working on design topics as well as front-end or back-end development.
I recently founded a "JUG-like" meetup in southern France called "Toulon Dev"","blueskyUsername":"cdetastes.bsky.social","company":"SCIAM","countryName":"France","firstName":"Clément","fullName":"Clément de Tastes","id":3653,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-0a33a77f-9982-4862-8714-ad3c32701c94.jpg","lastName":"de Tastes","linkedInUsername":"clement-de-tastes","mastodonUsername":"","twitterHandle":""}],"summary":"Interactive Java session exploring tricky language subtleties through short code snippets. Participants will predict outputs together, revisit core concepts, challenge assumptions, and deepen their understanding of how Java actually behaves in practice.","timeSlots":[],"title":"What Does it Print? The Java Quiz (Hard Edition 🤯)","totalFavourites":6,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Spring Boot 4.x and Java 25+ have arrived, making it an exciting time to be a Java developer! Join me, Josh Long (@starbuxman), as we dive into the future of Spring Boot with Java 25. We'll learn how to scale your applications and codebases effortlessly, explore the robust Spring Boot ecosystem featuring AI, modularity, seamless data access, security, and cutting-edge production optimizations like Project Loom's virtual threads, Project Leyden, GraalVM, and more. Let's explore the latest-and-greatest in Spring Boot to build faster, more scalable, more efficient, more modular, more secure, and more intelligent systems and services.
","fieldGroup":null,"fieldGroupValues":[],"id":19733,"keywords":[{"name":"graalvm"},{"name":"java"},{"name":"virtual threads"},{"name":"spring boot"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Josh (@starbuxman) has been the first Spring Developer Advocate since 2010. Josh is a Java Champion, author of 7 books (including "Reactive Spring") and numerous best-selling video training (including "Building Microservices with Spring Boot Livelessons"with Spring Boot co-founder Phil Webb), and an open-source contributor (Spring Boot, Spring Integration, Axon, Spring Cloud, Activiti, Vaadin, etc), a Youtuber (Coffee + Software with Josh Long as well as my Spring Tips series ), and a podcaster ("A Bootiful Podcast").
","blueskyUsername":"starbuxman.bsky.social","company":"the Spring team","countryName":"United States of America","firstName":"Josh","fullName":"Josh Long","id":14654,"imageUrl":"https://lh3.googleusercontent.com/a-/AOh14GhTAlrBfIHBoWpP1mfcKm-kRON4bdBF_LccGehD=s96-c","lastName":"Long","linkedInUsername":"joshlong","mastodonUsername":"","twitterHandle":"starbuxman"}],"summary":"Explore Spring Boot 4.x with Java 25+, covering scalable, modular, secure, and intelligent applications. Learn about the Spring ecosystem, AI integration, seamless data access, and production optimizations including virtual threads, Project Leyden, GraalVM, and more for building faster, more efficient systems.","timeSlots":[],"title":"Bootiful Spring Boot 4","totalFavourites":26,"track":{"id":1252,"name":"Architecture & Modernization","description":"Designing for agentic workflows, event-driven systems, and legacy transformation.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-blockchain.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Destructoring—the breakdown of an object into its constituent parts—is key to Java's ability to reliably maintain encapsulation and invariants. We resolve this apparent paradox in the presentation by looking at past and future language and platform features in the JDK.
Classic Java serialization breaks encapsulation: it directly accesses private fields, thereby bypassing constructors and the checks anchored there. The reason for this is that deconstruction and reconstruction have not been first-class citizens in the JDK until now.
This is precisely where current work within the JDK Project Amber comes in – under the working title “Derived Record/Class Creation.” This approach allows us to explicitly specify how objects are deconstructed and reconstructed. This will not only make serialization more secure and robust in the future, but we will also gain “withers” (targeted, immutable copies with individual modified components) for records and pattern matching, which will be extended beyond pure record patterns to normal classes.
Given these circumstances, it is worth taking a closer look: What is already there, what comes next – and what does this mean for our daily Java practice?","fieldGroup":null,"fieldGroupValues":[],"id":7012,"keywords":[{"name":"encapsulation"},{"name":"serialization"},{"name":"destructuring"},{"name":"records"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Richard Gross is an IT archaeologist at MaibornWolff with more than ten years of modernization experience. His focus is on hexagonal architectures, hypermedia APIs, TestDSLs and the expressive and unambiguous modelling of the domain as code. He enjoys mastering TDD, BDD, DDD, decoupled design and even practices that don't include two D's. He also shaped the open source project CodeCharta, which lets even non-developers grasp the quality of their software.","blueskyUsername":"richargh.de","company":"MaibornWolff GmbH","countryName":"Germany","firstName":"Richard","fullName":"Richard Gross","id":5839,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-7f802c4e-9a8f-467c-99d9-7d96c94b9ebc.jpg","lastName":"Gross","linkedInUsername":"richargh","mastodonUsername":"https://mas.to/richargh","twitterHandle":""}],"summary":"Project Amber’s “Derived Record/Class Creation” aims to make object deconstruction and reconstruction first-class in Java, fixing serialization’s encapsulation issues. It enables safer, more robust serialization, plus withers and broader pattern matching support for records and normal classes.","timeSlots":[],"title":"Destructoring is the future of Javas Encapsulation","totalFavourites":13,"track":{"id":2754,"name":"Java Language & Platform","description":"Project Loom, Leyden, Valhalla, Panama, and the ever evolving JVM.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-coffee-mug.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"Has this ever happened to you: You just spent €50,000 on a humanoid
robot, but it's not very fun at parties, so you have to train it to
dance using reinforcement learning in a simulation environment on your
€100,000 GPU cluster? What do you mean "NO"?
In this talk, Thomas Endres and David Winderl will teach you how you can
train your humanoid robot to dance using Reinforcement Learning (RL) in
simulation environments. They're gonna kick things off with an
introduction into RL, covering the basics of policies and rewards. After
that, they’ll take their first steps in simulation environments with the
introduction of NVIDIA Isaac Sim and the Unitree RL Lab. Our speakers
will run through the steps of shaping a walking policy, one hilarious
fall at a time. Finally, they're going to shake things up with a dancing
policy: You will learn how you can retarget human motion from videos to
the Unitree G1 robot and train an imitation policy that can dance just
as poorly as you.
For the grande finale, our humanoid robot will close the session with a
dancing performance, not missing out on a single step.","fieldGroup":null,"fieldGroupValues":[],"id":50103,"keywords":[{"name":"imitation"},{"name":"simulation"},{"name":"humanoid"},{"name":"reinforcement"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"Thomas Endres is a Managing Partner for TNG Technology Consulting in Munich. Besides his normal work for the company and the management of the customers he is creating various prototypes, including real-time deepfakes, an AI for generating presentations or various robotics showcases. He works on applications in the field of AI, robotics, AR/VR and teleoperation. Furthermore, he is involved in Open Source projects.
Thomas studied computer science at TU Munich and is a passionate software developer. As an Intel Software Innovator and Black Belt, he presents new technologies such as AI, AR/VR, and robotics worldwide. For this, he has received awards including a JavaOne Rockstar Award and several Best Speaker Awards.","blueskyUsername":"","company":"TNG Technology Consulting GmbH","countryName":"Germany","firstName":"Thomas","fullName":"Thomas Endres","id":50052,"imageUrl":"https://lh3.googleusercontent.com/a/AATXAJxWxubT7g3QJfDNjLG2FdVGAtl_5CZk3Jf_ukgT=s96-c","lastName":"Endres","linkedInUsername":"thomasendres","mastodonUsername":"","twitterHandle":""}],"summary":"A humorous talk on training a humanoid robot to walk and dance using reinforcement learning in simulation. It introduces RL basics, NVIDIA Isaac Sim, Unitree RL Lab, motion retargeting from video, and imitation learning, ending with a live dancing performance by the robot.","timeSlots":[],"title":"Let's dance! – Teaching your robot some moves with reinforcement learning","totalFavourites":null,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"The idea stems from https://xkcd.com/2347/ and Mark Rendle’s talk at Devoxx.be 2024.
Most people are unaware of the hassles we face (or inflict upon ourselves) regarding legacy systems, technical choices, or our own organizational structures.
The goal is to have a laugh about our points of contention (Lombok? Reactive? Date formats?), our nemeses (printers, CORS), language quirks—including the fact that you can run *Doom* inside the TypeScript type system—why no one truly understands C++, wacky HTTP status codes, centering `div`s (of course), the Y2K non-event versus the real Y2K38 problem, RGB vs. CMYK, April Fools' RFCs, web page and file encoding issues, consistently botched Excel integrations, business-dictated file naming conventions, the headache of IEEE 754 rounding, the "ultimate" email validation regex, hidden (or missing) CPU instructions, the shift toward inclusivity since 2020 (dropping "master/slave" terminology...), SOA, the long-awaited WASM, corporate timesheets, and—last but not least—the crucial question: *how do you exit vi?*","fieldGroup":null,"fieldGroupValues":[],"id":10031,"keywords":[{"name":"programming"},{"name":"quirks"},{"name":"humor"},{"name":"legacy"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#FF0000","description":null,"dinerInvite":true,"duration":50,"freeTickets":2,"id":951,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Conference","pause":false,"scheduleInSmallRoom":false,"slug":"conf50"},"speakers":[{"anonymizedBio":null,"bio":"A backend Java cowboy and impossible algorithms adept, proficient in Kafka, K8s, and Mongo, I specialize in "IT battlefield medicine" and hotfixing systems in the most unlikely places. I’ve held various roles across IT consultancies, startups, and freelance gigs, and even taught at an engineering school. I’m a huge electronic music collector and have even run a music label.
Early in my career, I worked on government contracts—which is how I ended up programming a good chunk of the embedded software for automated speed cameras. After Covid, I handled orders and logistics for Cultura, followed by a nuclear-sector project at EDF. I’m now working into occult stuff for Betclic.
I appreciate people who write specs and those who aren't afraid to pop the hood. I love trolling the corporate "NPCs." I coined the phrase "coding like a bear" to describe being "in the zone," and I make a habit of swearing at my AI assistant in English.
","blueskyUsername":"fabmars.bsky.social","company":"Nonstatic","countryName":"France","firstName":"Fabien","fullName":"Fabien Marsaud","id":10164,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-d8cda6a6-3355-41db-9892-3d55de893bed.jpg","lastName":"Marsaud","linkedInUsername":"fabmars","mastodonUsername":null,"twitterHandle":null}],"summary":"A humorous talk about the absurdities of software engineering: legacy systems, language quirks, bad integrations, confusing standards, and endless debates like Lombok vs. reactive, CORS, date formats, and vi. It riffs on technical pain points, from Y2K38 and IEEE 754 to WASM, Excel, and bizarre HTTP codes.","timeSlots":[],"title":"The Dark Secrets of Computer Science","totalFavourites":15,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"Java developers have long relied on JNI or manually maintained native bindings to access GPU libraries such as CUDA, cuBLAS, cuFFT, and cuDNN. While these approaches unlock GPU performance, they also introduce complexity, reduce portability, and make Java applications harder to maintain.
This Tools in Action session introduces TornadoVM's new Hybrid API, a new programming model that allows Java developers to seamlessly combine GPU kernels written in Java with direct calls to the NVIDIA software ecosystem. Instead of switching between Java and C++, developers can write high-level Java code while transparently invoking NVIDIA's highly optimized AI and HPC libraries from the same application.
Through live coding demonstrations, attendees will learn how to:
- Write GPU kernels directly in Java.
- Access CUDA runtime APIs from Java.
- Invoke cuBLAS and cuFFT without writing JNI code.
- Combine custom GPU kernels with vendor-optimized libraries for maximum performance.
- Incrementally accelerate existing Java applications without changing their preferred JDK.
Whether you're building AI applications, scientific simulations, financial analytics, or data-processing pipelines, this session will demonstrate how modern Java can access GPU hardware and NVIDIA libraries without sacrificing developer productivity or leaving the JVM.","fieldGroup":null,"fieldGroupValues":[],"id":19732,"keywords":[{"name":"java"},{"name":"gpu"},{"name":"nvidia"},{"name":"jni"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"Dr. Athanasios Stratikopoulos (male) is a Research Fellow at the University of Manchester with specialization on heterogeneous architectures and reconfigurable accelerators. He has authored more than 20 research articles in the field of hardware acceleration, system software and programming languages. Currently his work involves heterogeneous architectures ranging from low-power devices to high-end cloud deployments. He is one of the lead developers of TornadoVM and has been part of the team for the last eight years. In addition to his core contributions to the system's technical development, Dr. Stratikopoulos leads the project's communication and dissemination efforts, helping to articulate its goals and advancements to both academic and industrial audiences through talks, documentation, and outreach activities.","blueskyUsername":"stratika.bsky.social","company":"TornadoVM","countryName":"United Kingdom","firstName":"Thanos","fullName":"Thanos Stratikopoulos","id":7649,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-1eb3708e-af15-4778-a560-15ebbea2b133.jpg","lastName":"Stratikopoulos","linkedInUsername":"stratika","mastodonUsername":"","twitterHandle":"thanos_str"},{"anonymizedBio":null,"bio":"Michalis Papadimitriou is a Senior Software Engineer at Neo4j and a Research Fellow at the University of Manchester. His core expertise includes open-source software development, hardware abstractions for high-level programming languages, compiler optimizations for GPU computing, and enabling large language model (LLM) inference on GPUs for the Java Virtual Machine (JVM).
He is focused on advancing high-performance AI and graph computing systems, with particular interest in GPU acceleration for machine learning workloads on the JVM through the TornadoVM framework, and continues to actively maintain the open-source GPULlama3.java project.
Before joining Neo4j, Michalis was a Research Fellow at the University of Manchester, where he contributed to the TornadoVM project as a Staff Software Engineer. Prior to that, he worked across a range of software stacks at Huawei Technologies and contributed to the open-source machine learning compiler Apache TVM while at OctoAI (formerly OctoML), which was later acquired by NVIDIA.","blueskyUsername":"mikepapadim.bsky.social","company":"Neo4j","countryName":"United Kingdom","firstName":"Michalis","fullName":"Michalis Papadimitriou","id":19768,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-85f7f25f-5f2c-4526-bed6-489be66ed5e1.jpg","lastName":"Papadimitriou","linkedInUsername":"michalis-papadimitriou","mastodonUsername":"","twitterHandle":"mikepapadim"},{"anonymizedBio":null,"bio":"I'm a Research Associate at the University of Manchester and lead developer of TornadoVM, an open-source framework that lets Java applications run on GPUs and other accelerators without leaving the JVM. My work sits at the intersection of managed runtimes and compilers - JIT-compiling Java bytecode into GPU code, and bringing modern hardware features like tensor cores, low-precision arithmetic and quantization into reach for Java developers.","blueskyUsername":"maryxek.bsky.social","company":"University of Manchester","countryName":"UK","firstName":"Mary","fullName":"Mary Xekalaki","id":49105,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-a82d2879-53f1-4330-b9cb-64cc383a1b02.jpg","lastName":"Xekalaki","linkedInUsername":"mary-xekalaki","mastodonUsername":"","twitterHandle":"MXekalaki"}],"summary":"TornadoVM’s Hybrid API lets Java developers combine GPU kernels written in Java with direct access to NVIDIA libraries like CUDA, cuBLAS, and cuFFT—without JNI. The session shows how to build high-performance AI/HPC apps in Java while preserving portability, productivity, and incremental acceleration within the JVM.","timeSlots":[],"title":"Tap into the NVIDIA Ecosystem from Java with TornadoVM's Hybrid API","totalFavourites":1,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"
The welcome opening keynote for Devoxx Belgium 2026 by the founder of Devoxx giving a general update on the conference.","fieldGroup":null,"fieldGroupValues":[],"id":31451,"keywords":[{"name":"belgium"},{"name":"conference"},{"name":"devoxx"},{"name":"keynote"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"#e89c1a","description":null,"dinerInvite":true,"duration":20,"freeTickets":2,"id":955,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Keynote","pause":false,"scheduleInSmallRoom":false,"slug":"key20"},"speakers":[{"anonymizedBio":null,"bio":"With three decades of experience in the Java ecosystem and as the founder of Devoxx, one of the world's most prestigious developer conferences, Stephan brings unparalleled expertise in developer tools and workflows. His AI journey began 7+ years ago with CNNs for basketball analysis, evolving into 5 years of intensive work with GPT models. As the creator of DevoxxGenie, he's been at the forefront of AI-assisted development, building real-world applications and helping thousands of developers level up their skills.","blueskyUsername":"","company":"Devoxx","countryName":"Belgium","firstName":"Stephan","fullName":"Stephan Janssen","id":30503,"imageUrl":"https://lh3.googleusercontent.com/a/ALm5wu39iQhWBfIgxSMBGJ-tLuCnWERa6Vp5RRnnq4exBw=s96-c","lastName":"Janssen","linkedInUsername":"","mastodonUsername":"","twitterHandle":""}],"summary":null,"timeSlots":[],"title":"Welcome to Devoxx","totalFavourites":13,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}},{"afterVideoURL":null,"audienceLevel":"INTERMEDIATE","description":"At some point, we’ve all had to (re)install our entire development environment after switching machines.
Whether by choice or simply during on-boarding at a new company, a new mission,...
We’ve also all spent far to many hours figuring out which tools to install just to submit a PR or MR to an open source project.
And in many other situations, we quickly find ourselves facing the nightmare of configuring tools and frameworks when all we really want to do is what we love: coding.
And let’s not even get started on version updates and inter-dependencies 😅...
Fortunately, there is a solution to all these problems: dev containers.
During this tools-in-action session, I invite you to explore them with me.
The only thing I allow myself to install on my machine: Docker!
Everything else will be handled through configuration so that tools, frameworks, version management, and dependencies become a non-event in your development workflow.
You’ll leave knowing how to use dev containers and how to build your entire development stack as code.
","fieldGroup":null,"fieldGroupValues":[],"id":5051,"keywords":[{"name":"dependencies"},{"name":"devcontainers"},{"name":"configuration"},{"name":"docker"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":"","description":null,"dinerInvite":true,"duration":30,"freeTickets":2,"id":957,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":1,"name":"Tools-in-Action","pause":false,"scheduleInSmallRoom":false,"slug":"tia30"},"speakers":[{"anonymizedBio":null,"bio":"During more than 20 years in the software development profession, I was able to confront myself with many different technologies. And it is this diversity that makes me continue to love our job so much.
For 5 years now, I have had the chance to evolve within OVHcloud as a developer advocate. This position allows me to discover new technologies or disciplines such as Artificial Intelligence or "native cloud" for example.
But my heart of old Java developer can not move too far from this technology which is the common thread of my many years as a developer. It is with humility that I share this experience but also the novelties of our ecosystem through my conferences and the TADx meetup co-created in Tours.","blueskyUsername":"wilda.bsky.social","company":"OVHcloud","countryName":"France","firstName":"Stéphane","fullName":"Stéphane Philippart","id":3925,"imageUrl":"https://devoxxian-image-thumbnails.s3-eu-west-1.amazonaws.com/profile-9d422934-263f-49b6-a07c-fbc56af630a9.jpg","lastName":"Philippart","linkedInUsername":"philippartstephane","mastodonUsername":null,"twitterHandle":"wildagsx"}],"summary":"Dev containers simplify development setup by codifying tools, frameworks, versions, and dependencies in configuration. With only Docker installed locally, developers can quickly recreate environments, avoid manual setup hassles, and manage updates consistently. This session shows how to use dev containers to build an entire development stack as code.","timeSlots":[],"title":"🧰 Dev Containers: the ultimate toolbox for developers?","totalFavourites":23,"track":{"id":2758,"name":"Mind the Geek","description":"Developer candy: stuff we want to know about but dont (generally) at work, Robotics, biological computing, cybernetics, AI, new toys, tomorrows world","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-robot-arm.png"}},{"afterVideoURL":null,"audienceLevel":"BEGINNER","description":"With Java, data can be loaded easily from different sources and displayed in a web UI with little effort. However, as soon as users need to search for specific records, applications often rely on complex filter forms. Modern LLMs offer an alternative: they can extract filter criteria from natural language and turn them into suitable database queries.
Using an example project built with Spring AI and Vaadin, we compare three possible approaches. With Function Calling, the model invokes tool functions in the application code and passes the required parameters. This is easy to understand, but in a simple implementation it quickly reaches its limits when handling complex requests involving negations, operators, or multiple values. Structured Output, on the other hand, directly creates a custom filter object and can represent more complex requests in a compact form. A hybrid approach combines both concepts by using a tool call whose parameters themselves contain a structured filter definition.
All approaches are tested with the same queries against local Ollama models. We compare which approach is more reliable, faster, and more cost-efficient, and examine where each of them reaches its limits.","fieldGroup":null,"fieldGroupValues":[],"id":50106,"keywords":[{"name":"java"},{"name":"llm"},{"name":"structuredoutput"},{"name":"filtering"}],"language":null,"podcastURL":null,"sessionType":{"countInQuota":true,"cssColor":null,"description":null,"dinerInvite":true,"duration":40,"freeTickets":1,"id":5151,"invitationOnly":false,"maxSpeakers":null,"minSpeakers":null,"name":"Lunch Talk","pause":false,"scheduleInSmallRoom":false,"slug":"lunch-talk"},"speakers":[{"anonymizedBio":null,"bio":"Sebastian has been a Java junkie for over two decades and fell in love with Vaadin during his university days. Since 2016, he has worked in various technical roles at Vaadin, helping customers build tailor-made web applications and leading workshops and training sessions. In 2024, he transitioned to looking after the community full-time, traveling from event to event as a speaker to share his knowledge and passion for Vaadin. When he’s not at a conference or experimenting with AI, he’s always up for a tech talk and a good cup of coffee.","blueskyUsername":"","company":"Vaadin","countryName":"Germany","firstName":"Sebastian","fullName":"Sebastian Kühnau","id":19764,"imageUrl":"https://lh3.googleusercontent.com/a/ACg8ocKJ7pHA4yXXBTPT1sa3oGVZ-ONpi-AtRLBtJIKuxKb5_sQOwm0=s96-c","lastName":"Kühnau","linkedInUsername":"sebastian-kuehnau","mastodonUsername":"","twitterHandle":""}],"summary":"The project compares three LLM approaches for natural-language database filtering in a Java Spring AI/Vaadin app: function calling, structured output, and a hybrid method. Tested with local Ollama models, it evaluates reliability, speed, cost, and limitations for handling complex filter requests.","timeSlots":[],"title":"From User Prompt to JPA Query: Comparing Tool Calling, Structured Output & Hybrid Approach","totalFavourites":null,"track":{"id":1251,"name":"Agentic Engineering & Tooling","description":"Multi-agent systems, tool-use, evals, and shipping AI beyond the demo.","imageURL":"https://s3.eu-west-1.amazonaws.com/devoxx.be/devoxxbe26-tracks/icon-skull-1.png"}}]