I build tooling around AI coding agents and the systems they run on, from smartwatch firmware to homelab servers.
- 🤖 Working daily with coding agents (Codex, Claude Code), and building on their protocols
- 🧪 Background in software testing: pytest, Playwright, Locust, regression suites
- 🔧 Languages: Go, Python, C, TypeScript; also learning Zig and Elixir
- 🐧 Linux, ARM64 homelab, and embedded work
Contributed a merged fix to prevent queued audio from being cut off when a speaker stream closes, with regression tests and API documentation.
C · Embedded systems · Audio · Regression testing
Fixed how the Codex integration counts human turns, so that explicit structured selections made by a person are recognised as human input.
AI agents · Codex · Workflow tooling
Added data classification at every model boundary, refusal of sensitive queries before they run, and protection of HR and personnel data.
LLM safety · Data governance · Python
Data classes at model boundaries → · Refused-before-run queries → · HR data protection →
