“0xCaml From a System Engineer's Point of View”
Dmitriy Kovalenko;
Talk
I do a lot of Rust, C, Zig and whatever else cursed systems programming languages are there. So recently 0xCaml popped out and I was def interested if I can finally use it. So I did some research on how it actually works comparing to native OCaml and other languages and try to build odiff (the fastest image comparison lib in the world) with 0xCaml. Here is what I learned.
“Build Language Models with Raven”
Thibaut Mattio;
Workshop
Discover how to build and train language models using OCaml with Raven, a modern scientific
computing ecosystem that brings JAX-like capabilities to functional programming. This hands-on
workshop introduces Raven's key libraries-Nx for tensor operations, Rune for automatic
differentiation and JIT compilation, and Kaun for neural network architectures-demonstrating how to
implement transformer-based models with type safety and competitive performance.
We'll explore Rune's effect-based autodiff system, implement attention mechanisms, and train a
small language model while leveraging OCaml's type system for production-ready ML systems.
“Frameworks: No, Libraries: Yes. Developing a Product in OCaml From Scratch”
Malcolm Matalka;
Talk
Terrateam is an open source infrastructure management product written in OCaml. In order to stay lean and flexible, Terrateam has developed almost all of its own frameworks and tooling. With a small team, the Terrateam product punches well above its weight compared to its competition. This talk will cover the choice to develop their own frameworks, the good, the bad, and how OCaml has been instrumental in being able to achieve so much despite being a two person company.
“Generating Static Websites the Functional Programming Way”
Xavier Van de Woestyne;
Talk
Static site generators like Jekyll, Zola, and Hugo are effective, but fall short for complex sites like personal encyclopedias, wikis, or sites with rich content. Features like backlinks and transclusions require richer context and efficient builds. This talk explores the theory behind static site generation (as a case of build systems) and shows how to implement advanced features inspired by Project Xanadu using YOCaml, a static site framework based on functional abstractions (such as strong profunctors) in OCaml.
“Slipshow: A Full-Featured Presentation Tool in OCaml”
Paul-Elliot Anglès d'Auriac;
Talk
Slipshow is a tool to create interactive presentations. Started as a JavaScript project, it is now fully written in OCaml and features a runtime engine, a compiler, a collaborative editing website, a VSCode extension, a standalone app...
This talk tells the story of how it is possible to develop and maintain such a challenging project as a single developer.
If you are reading this far into the abstract, I suspect you are going to FUN OCaml? Are you giving a talk? If so, consider (re)writing your presentation in Slipshow!
“Train a Reinforcement Learning Agent for the Box-Pushing Game Sokoban in OCANNL or Raven”
Lukasz Stafiniak;
Workshop
I will give a very brief intro to Reinforcement Learning (RL), covering just the REINFORCE algorithm. We will build an RL harness and an environment for the Sokoban game (cheating allowed). We will program and train models using a convolutional neural net or an axial-attention transformer. While the nets are training, I will introduce the Group Relative Policy Optimization (GRPO) algorithm and discuss how our agent setup relates to post-training of language models on solving problems. Time permitting, we would then implement GRPO and compare its performance with REINFORCE.