An airhacks.fm conversation with Gil Tene (@giltene) about: starting with hacking adventure games on a VAX-11/780 as a teenager, building computers and making money in high school, providing access to Usenet, early programming experiences with Pascal and C/C++, moving to Silicon Valley in 1994 and witnessing the rise of Java, working on fault-tolerant computer systems at Stratus Computer, co-founding Azul Systems and developing the Vega appliances to virtualize Java applications, the technical details of how Vega appliances worked by running JVMs on specialized hardware, the evolution of Azul to focus on pure software solutions such as Zing and supporting openJDK, Gil's continued involvement in coding and maintaining open-source libraries

Gil Tene on twitter: @giltene

An airhacks.fm conversation with Dr. Zoran Sevarac (@zsevarac) about: Zoran previously on airhacks.fm: "#169 Deep Learning with Modern Java Code", discussion about the latest updates and features in DeepNetts, a full-stack Java AI platform, University of Minnesota's drug testing application using DeepNetts, Jefferson Lab's particle research using DeepNetts Community Edition, including GPU support for faster inference using jcuda, TensorFlow compatibility, and simplified AI integration with JSR-381, real-world applications of DeepNetts in drug testing and particle research, challenges and considerations for using GPUs in serverless environments, the potential of Apple's M-series chips for machine learning, exploring Project Babylon and Code Reflection in Java, using Panama and jextract for native library bindings, the importance of having developer tools and an IDE for building AI models, plans for integrating large language models into DeepNetts, the advantages of a pure Java solution for AI in enterprise applications, and the bright future of Java in the AI ecosystem, Deep Nets 3.1.0 release with GPU support

Dr. Zoran Sevarac on twitter: @zsevarac

An airhacks.fm conversation with Jonathan Schneider (@jon_k_schneider) about: from Pentium 2 machine and a rural high school to becoming a Java refactoring entrepreneur, self-taught C++ in high school, officer in the U.S. Army and deployed to Iraq and Afghanistan, worked on Java projects at an insurance company between deployments, joined Netflix to work on engineering tools, challenges of migrating Java versions and libraries in a freedom and responsibility culture, started the OpenRewrite project at Netflix for automated refactoring and code migration, founded the micrometer metrics instrumentation project at Pivotal, challenges of introducing automated pull requests in enterprise environments, rebooted OpenRewrite while working with Gradle's Hans Dockter, founded Moderne to commercialize OpenRewrite for large-scale enterprise refactoring, the origin of the Moderne name and its Art Deco roots, OpenRewrite's Maven and Gradle plugins for refactoring and styling using a visitor pattern on an enriched AST, Moderne's enterprise offering for large-scale refactoring and impact analysis, potential integration with large language models and retrieval-augmented generation for code optimization

Jonathan Schneider on twitter: @jon_k_schneider

An airhacks.fm conversation with Nicolai Parlog (@nipafx) about: Nicolai previously on "#206 Java 19: Millions of Threads in No Time", discussion about the underscore feature in Java 22 and its importance in pattern matching, using the underscore for unused lambda parameters and deconstruction of records, avoiding default branches when switching over sealed types, the deprecation and removal of underscore as a regular variable name, the foresight of the Java community in making underscore unusable, the simplicity of installing Java compared to other languages, the need for a minimalistic Java build tool for better developer experience, SdkMan, the bld tool as an example of a pure Java build tool, the process of contributing to OpenJDK and the importance of starting with a problem statement, the distinction between Java specifications and implementations, the concept of long-term support (LTS) in Java and its relation to vendors, the importance of using the right terminology to avoid misunderstandings in the Java ecosystem

Nicolai Parlog on twitter: @nipafx

An airhacks.fm conversation with Dimitris Andreadis (@dandreadis) about: Dimitris appeared previously on "#64 Quarkus 1.0 and SpringBoot", discussion about integrating AI language models (LLMs) with Java applications using quarkus and langchain4j, OpenShift AI, the benefits of using Quarkus for AI integration, Drools and ML, the potential of using AI for rule engines and decision making, the challenges of handling state and context with LLMs, InfiniSpan and vector databases, the role of vector stores and embeddings for semantic search, the advantages of Java for enterprise applications and maintenance, the potential of using AI models natively with GraalVM, the importance of tools functionality for LLMs to call Java methods, the excitement around AI in the Java community, the future trajectory of tighter integration between Java and AI models, the potential of using AI for code generation and intelligent developer tooling

Dimitris Andreadis on twitter: @dandreadis

An airhacks.fm conversation with Erik C. Thauvin (@ethauvin) about: early computer experiences with Logo and Tandy Model 3, writing horse race handicapping software as a kid, working at Apple at 16 writing resource editor for Mac, starting consulting firm and building custom software, attending Sun Tech Days to learn about Palm Pilot and Java, writing linkblog with Tomcat and JSP, creating popular linkblog with 8 million monthly views, converting projects to Rife and BLD frameworks, motivations for writing software he needs, Erik's blog: erik.thauvin.net

Erik C. Thauvin on twitter: @ethauvin

An airhacks.fm conversation with Paul Sandoz (@paulsandoz) about: Paul previously appeared on "#277 Project Babylon", article about Automatic differentiation of Java code using code reflection, using Project Babylon for automatic differentiation, emulating C# LINQ in Java, writing Triton programs in Java for GPU execution, using Project Panama for native interoperability with C/C++ libraries, JExtract tool for generating Java bindings from C headers, memory management with Panama memory segments and arenas, advantages of Panama over JNI, integrating Java with ONNX models, persistent memory and object storage

Paul Sandoz on twitter: @paulsandoz

An airhacks.fm conversation with Dmytro Liubarskyi (@langchain4j) about: continuous Windows 95 re-installation on Pentium 2 then Pentium 3, early interest in J2ME development and websites, transition to Java and enterprise software development, motivation behind creating langchain4j, integration with embedding models, vector databases, and ONNX Runtime, langchain4j core abstracts: language models, chat memory, AI services, tools, langchain4j, onnx.ai, Dynamic tools with GraalVM, Enterprise use cases and integration with Java stacks, OpenAI, Anthropic, Cohere, Hugging Face, Bedrock, Olama, Gemini

Dmytro Liubarskyi on twitter: @langchain4j

An airhacks.fm conversation with Geert Bevin (@gbevin) about: Yamaha DX7 - Geerts first synthesizer, Classic Minimoog analog synthesizer by Bob Moog, First synthesizer was a Yamaha DX7 successor, Early music production using Amiga computer and MIDI, iOS and macOS development with focus on audio DSP and synthesis, Returning to Java development after years away, high productivity with Java, New Java release cadence and experimental features, Rife web framework and bidirectional template engine, BLD build tool and philosophy, Kotlin Multiplatform capabilities, SwiftUI DX compared to Java and IntelliJ, Moog Music website, Jean Michel Jarre, Geert personal website, Rife 2 GitHub project, Bld GitHub project Geeert's website: uwyn.com

Geert Bevin on twitter: @gbevin

An airhacks.fm conversation with Jose Paumard (@JosePaumard) about: discussion about Object-oriented programming vs data-oriented programming in Java, using Java record classes, hexagonal architectures, considerations for decoupling and abstractions, the advantages of polymorphism and transparent persistence, the pitfalls of inheritance and abstraction, the importance of naming and cohesion sealed interfaces and switch expressions to implement business logic outside of data classes, using sealed types and switch statements, advantages of better separation of concerns and ease of removing unused code, data-oriented programming with NoSQL, the Citroen 2CV (Duck) Car

Jose Paumard on twitter: @JosePaumard

An airhacks.fm conversation with Juan Fumero (@snatverk) about: Juan previously appeared in the episode "#250 FPGAs, GPUs or Data Science with Java", using Tornado to run Java programs on GPUs/accelerators, integrating AI models with Java applications, potential of using Tornado and Project Babylon together, discussion around tensor types in Java, Paul Sandoz appeared in the episode "#277 Project Babylon", Heterogeneous Accelerator Toolkit by Gary Frost, TornadoVM and LLama port, Hybrid API for Deep Learning acceleration and the new Panama-based types: TornadoVM talk at JVMLS'23, TornadoVM 1.0 Release notes, Alfonso Peterssen ported llama to Java, Initial Java port from the GraalVM team, Java / AI startup: paravox.ai

Juan Fumero on twitter: @snatverk