I design how people and AI think together, then build software with that method.
Now: contributing to an iOS astrology app from a fork: a lunar-first Home face, engine accuracy fixes, and the astrologer's tooling. Western, Vedic and Human Design, computed rather than guessed.
How I work
- One concern per branch, named for the trunk it targets. Drafts until they're ready.
- Every PR carries its receipts: what ran, on what, and what didn't.
- An independent review before anything is handed over.
- Calls that belong to someone else go in as questions, not assumptions.
- Errors get named before they get fixed, and the fix becomes a rule.
- No personal data in code, docs or history.
Building: sia-research-engine (multi-agent deep research) 路 claude-harness (Claude Code environments) 路 WINENV (a workstation as code) 路 luma-production-engine (AI video production)