I write a lot of code in a lot of languages. Mostly functional ones (Standard ML, OCaml, Haskell, Clojure, Scala, Racket, Lisp, Elixir, F#), but also Rust, C, Python, Go, and TypeScript.
- Languages and parsers: Lisp/Scheme interpreters, parser combinators, regex engines, esolangs, a small SQL engine.
- Games: libraries and prototypes with raylib (C), phaser (TS/JS), love2d, and a few hand-rolled engines (ECS, pathfinding, collision, FOV).
- Machine learning: deep-learning experiments and a multi-agent coding agent.
- Cryptography and Bitcoin: hashes, ciphers, elliptic curves, signatures, Bitcoin/Ethereum primitives.
- Systems: codecs, compression, an acoustic modem (data over sound), tooling in Rust and C.
- Classics and writing: interlinear Latin, Greek, and Old English, plus longform writing in TeX.
- forge — an autonomous coding agent on a multi-agent LangGraph orchestrator.
- sonance — an adaptive acoustic modem in Rust (OFDM, LDPC).
- raydial and raypals — raylib libraries for dialogue boxes and sprites (C).
- chinalang — an esoteric language with Lisp-style macros; tree-walking interpreter in Go.
I maintain about 100 pure Standard ML libraries. They cover crypto and Bitcoin (sml-secp256k1, sml-ed25519, sml-sha3), parsers and serialization (sml-json, sml-cbor, sml-parsec), graphics (sml-glm, sml-image, sml-raster), data structures (sml-graph, sml-rope, sml-crdt), numerics (sml-matrix, sml-rational), and a web stack (sml-http, sml-router, sml-jwt, sml-ws).
They have no external dependencies, run on both MLton and Poly/ML, and are tested against the relevant RFC/spec vectors.