Cert-grade systems engineering. Verifiable where it matters; functional where it helps.
Mostly heads-down; occasional contributions to projects we depend on or admire.
| Modern | Foundational | |
|---|---|---|
| Systems | Rust, Zig, Odin, Hare | C, C++, Ada, Forth, Assembly |
| Functional | Haskell, OCaml, Scala, F#, Erlang/Elixir, Lean 4 | Lisp, Scheme, ML, Miranda, Lambda calculus |
Strong opinions on systems: ownership, type-level invariants, allocator discipline, compile-time composition. The closer to the metal the work runs, the more interesting it gets.
Strong opinions on functional: effect tracking, algebraic data types, supervision discipline, the bias toward total functions over partial ones. The right type makes the wrong state hard to spell.
| No swallowed failures | Errors propagate or get explicit rationale |
| Illegal states unrepresentable | Encode invariants at the type level |
| Pure cores, effectful edges | Effects tracked; total where the type permits |
| Exhaustive pattern matching | No catch-alls that swallow new cases |
| No accidental quadratics | Complexity claimed up front; WCET when deployed |
| Strict-by-default static analysis | Explicit allowance ledger for every exception |
| Qualified toolchains, MISRA-grade source | Where the deployment context calls for it |
| Mutation-tested coverage | Tests that don't kill mutations get rewritten |
| Supply-chain discipline | cargo-deny for Rust; equivalents elsewhere |
| Independent peer review | Before any change ships |