airhacks.fm podcast with adam bien

An airhacks.fm conversation with Stephan Janssen (@Stephan007) about: Devoxx Belgium conference selling out in seconds with 7000+ page views, ticket coordination strategies by companies, VAT validation issues with European services, conference featuring largest Java AI content globally, AI and LLM focus with dedicated GenAI track, MCP and AI agents presentations, Java on GPU with tornadovm and Babylon project, quarkus AI agents for langchain4j, vibes coding experiences with React and Angular, importance of specifications for AI code generation, using multiple LLMs for review (Claude, GPT, DeepSeek), file system RAG replacing vector databases, MCP as evolution of RAG, Java as optimal language for LLM code generation due to type safety and standards, bce pattern and MicroProfile for consistent code structure, package-info files improving LLM context understanding, junior developers benefiting from AI as universal tutor, CSS generation without manual coding, dark/light theme support using CSS variables, enterprise challenges with dependency management, DevoxxGenie plugin popularity in enterprises with restricted AI access, open source contribution challenges, local LLMs becoming powerful for enterprise inference, TornadoVM Metal support enabling local inference, comparison with GraalVM inference performance, AI disrupting video production and creative industries, importance of prompt engineering skills, renaissance of development through AI empowerment Stephan Janssen on twitter: @Stephan007

An airhacks.fm conversation with Antonio Goncalves (@agoncal) about: journey from Java Champion to Principal Software Engineer at Microsoft focusing on AI, the evolution from Java EE standards to modern AI development, writing technical books with LLM assistance, langchain4j as a Java SDK for LLMs providing abstraction over different AI providers, the importance of Java standards and patterns for LLM code generation, Boundary Control Entity (BCE / ECB) pattern recognition by LLMs, quarkus integration with LangChain4J enabling dependency injection and multi-tenancy, MCP (Model Context Protocol) as a new standard potentially replacing some RAG use cases, enterprise AI adoption using Azure AI Foundry and AWS Bedrock, model routers for optimal LLM selection based on prompt complexity, the future of small specialized models versus large general models, tornadovm enabling Java execution on GPUs with 6x performance improvements, GraalVM native compilation for LLM applications, the resurgence of Java EE patterns in the age of AI, using prompts as documentation in READMEs and JavaDocs, the advantage of type-safe languages like Java for LLM understanding, Microsoft's contribution to open source AI projects including LangChain4J, teaching new developers with AI assistance and the importance of curiosity, CERN's particle accelerator and its use of Java, the comparison between old "hallucinating architects" and modern LLM hallucinations, writing books about AI using AI tools for assistance, the structure of the Understanding LangChain4j book covering models RAG tools and MCP, enterprise requirements for data privacy and model training restrictions Antonio Goncalves on twitter: @agoncal

An airhacks.fm conversation with Michalis Papadimitriou (@mikepapadim) about: GPU acceleration for LLMs in Java using tornadovm, evolution from CPU-bound SIMD optimizations to GPU memory management, Alfonso's original Java port of llama.cpp using SIMD and Panama Vector API achieving 10 tokens per second, TornadoVM's initial hybrid approach combining CPU vector operations with GPU matrix multiplications, memory-bound nature of LLM inference versus compute-bound traditional workloads, introduction of persist and consume API to keep data on GPU between operations, reduction of host-GPU data transfers for improved performance, comparison with native CUDA implementations and optimization strategies, JIT compilation of kernels versus static optimization in frameworks like tensorrt, using LLMs like Claude to optimize GPU kernels, building MCP servers for automated kernel optimization, European Space Agency using TornadoVM in production for simulations, upcoming Metal backend support for Apple Silicon within 6-7 months, planned support for additional models including Mistral and gemma, potential for distributed inference across multiple GPUs, comparison with python and C++ implementations achieving near-native performance, modular architecture supporting OpenCL PTX and future hardware accelerators, challenges of new GPU hardware vendors like tenstorrent focusing on software ecosystem, planned quarkus and langchain4j integration demonstrations Michalis Papadimitriou on twitter: @mikepapadim

In this engaging conversation, Adam Dudczak, a seasoned software engineer and tech enthusiast, reminisces about his early coding adventures on the Commodore 64. He shares fascinating tales of the demo scene and how he transitioned through various programming languages. Dudczak reveals insights from his work at Allegro, discussing its e-commerce dominance and the tech innovations behind it. He also dives into cloud strategies, contrasting Spring with Quarkus, while offering tips on optimizing public cloud costs—making this chat packed with both nostalgia and modern tech wisdom!

Maurice Naftalin, an expert in Java generics and author of 'Java Generics and Collections', dives into the fascinating history of Java generics. He discusses the impactful Pizza paper by Phil Wadler, the complications of type erasure, and the evolution of Java collections. Maurice explains covariance and contravariance while highlighting the importance of wildcards. He also touches on the significance of unchecked warnings, the use of generics for better plugin systems, and gives insights into his upcoming book with future updates. A must-listen for Java enthusiasts!

Join Jonathan Ellis, a developer at Brokk and a coding innovator, as he dives into a multitude of fascinating topics. He explores Brokk’s intuitive design for AI-driven development and its advantages over traditional tools like Cursor. The conversation covers real-time feedback enhancements, in-memory parser optimizations, and innovative coding techniques like hierarchical embeddings. Jonathan also discusses the role of AI in efficient coding practices, subscription models, and the importance of structured codebases, making this chat a treasure trove for developers!

Fabio Niephaus, an Oracle expert focused on GraalVM and WebAssembly, dives into the exciting world of polyglot programming. He discusses how GraalVM's capabilities enhance Java apps by integrating languages like Python and JavaScript. Explore the innovative WebImage feature that compiles Java bytecode to WebAssembly, enabling seamless web execution. Fabio also highlights WebAssembly's role in secure code execution and its potential for efficient edge computing, promoting new project opportunities and innovations within the Java ecosystem.

Maurice Naftalin, co-author of 'Java Generics and Collections' and a seasoned Java programmer, shares fascinating insights from his experiences in the tech world. He discusses the evolution of integrated development environments (IDEs) and emphasizes the advantages of Visual Studio Code in a polyglot landscape. The conversation also dives into Java's historical changes, the importance of generics, and the elegance of method references. Naftalin's stories from rural Scotland and his journey through the transformations in programming paint a vivid picture of the industry.

Michalis Papadimitriou, an expert in GPU acceleration and compiler optimizations for AI and Java, shares fascinating insights from his journey in tech. He discusses how he achieved up to 20x speedups in Java applications by leveraging OpenCL and TornadoVM. Hear about his work at Huawei and how he is optimizing AI frameworks like Llama 3, emphasizing the importance of standardizing ML model formats. With a focus on enhancing GPU processing in Java, he highlights kernel fusion techniques and the exciting potential of Graal VM in the modern developer landscape.

Jonathan Ellis, the creator of Brokk, an AI tool for writing code, dives deep into AI's role in enhancing programmer productivity. He discusses the evolution of coding practices, comparing the productivity of younger versus older programmers. The conversation reveals the unique features of Brokk, including context sculpting for improved AI interactions and its open-source nature. Ellis also highlights challenges within large codebases and the use of AI for code reviews and technical debt management, showcasing a shift in mindset among developers.