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 type theory: Lisp/Scheme interpreters, parser combinators, regex engines, esolangs, a small SQL engine, a self-hosting dependently-typed language, and machine-checked proofs in Lean 4.
• Machine learning: GPT-2 inference written from scratch (including in pure Standard ML), reverse-mode autodiff, and a multi-agent coding agent.
• Cryptography and Bitcoin: hashes, ciphers, AEAD, elliptic curves, signatures, and Bitcoin/Ethereum primitives.
• Systems: codecs, compression, a hypergraph rewriting engine, an acoustic modem (data over sound), and tooling in Rust and C.
• Games: libraries and prototypes with raylib (C), phaser (TS/JS), love2d, and a few hand-rolled engines (ECS, pathfinding, collision, FOV).
• 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, built around outcome-honesty invariants and baseline-aware merge gates.
• blight — a self-hosting, dependently-typed language with a tiny trusted kernel (cubical + quantitative types + effects), in Rust.
• multiway — a multiway hypergraph rewriting engine with e-graph-style state sharing; zero dependencies, deterministic to the byte.
• sonance — an adaptive acoustic modem in Rust (OFDM, LDPC, fountain codes).
• raydial and raypals — raylib libraries for dialogue boxes and sprites (C).
• chinalang — an esoteric language with Chinese keywords and Lisp-style macros; tree-walking interpreter in Go.
• lean-* — verified data structures and algorithms in Lean 4 (red-black trees, ROBDDs, union-find, DFA/NFA, DPLL), each with machine-checked proofs.

I maintain more than 360 pure Standard ML libraries. They have no external dependencies, run on both MLton and Poly/ML, and are tested against the relevant RFC/spec vectors — many are byte-identical across the two compilers.

A rough map of the collection:

• Math, numeric & scientific: sml-matrix, sml-rational, sml-tensor, sml-pde, sml-kalman
• Cryptography & Bitcoin: sml-secp256k1, sml-ed25519, sml-aes, sml-aead, sml-rsa, sml-merkle
• Encoding & serialization: sml-json, sml-cbor, sml-asn1, plus codecs and compression
• Web & networking: sml-http, sml-router, sml-jwt, sml-ws, sml-tls
• Parsing & languages: sml-parsec, and a Standard ML toolchain written in Standard ML — sml-mlast, sml-elab, sml-lsp, sml-lint, sml-fmt, sml-pkg
• Logic & proofs: sml-sat, sml-smt, sml-bdd, sml-hm
• Machine learning: GPT-2 in pure SML — sml-gpt, sml-transformer, sml-bytebpe, sml-safetensors, sml-nn
• Data structures: sml-graph, sml-rope, sml-crdt
• Graphics & UI: a self-drawn immediate-mode GUI (sml-ui) that opens a real desktop window (sml-window)

…plus games, databases, information retrieval, audio, and more.

awesome-standard-ml is a curated index of the whole ecosystem.