My name is Cat and I'm a lead developer from the UK.
A few of my specialities are:
- Rust, C++, C#, Python, Haskell (but I'm known for being able to use just about any language :P)
- High performance applications such as 3d graphics, GIS and data processing
- GPU programming, from CUDA ray tracing to real-time renderers
- Building end-to-end software and features that actually work and solve problems
- Cloud and microservices architectures with Kubernetes, Docker and Azure
- Building software with AI, building AI tooling and MCP servers, and lately learning to train my own language models from scratch
I've been into neural nets since well before transformers: neuron is a feed-forward net trainer I wrote from scratch in Haskell back in 2016.
Some things I've made that I'm fond of:
- CatApp: the organiser I use to keep my life and projects on track, with AI agents posting updates over MCP and an assistant you can talk to
- VolumeRender: a sparse voxel octree ray tracer in CUDA that handles refraction through heterogeneous volumes (my dissertation)
- better_tactics: a chess tactics trainer with spaced repetition
- instructive-chess: R&D in turning narrated chess videos into data an AI can learn from. Computer vision reads every game, clock and rating off YouTube, a benchmark checks each one against the real game, and the masters' commentary is kept timed to their moves. It's a searchable database of instructive games for now, and next up is an AI opening and position instructor built on it
- dreamfield: a PS1-style first-person RPG in Rust, with its own engine written by hand before AI tools. I spent a long time researching how the PlayStation actually drew things and reproducing it faithfully: vertices snapping to the pixel grid, affine texture warping, the PS1 GPU's own dither matrix, and a composite video pass that bleeds colour like an old CRT
And from further back, coment: a header-only C++ entity-component-system library that picked up 54 stars.
See https://cat.bio/ for my full profile 👍