Research-oriented engineering at the boundary of symbolic AI and production systems. Includes work on OpenCog's Hyperon project — porting a genetic programming system to a new language runtime.
Supervises local dev servers behind stable proxy ports, so a project's URL never changes even as the process behind it restarts. Idle-kills on real traffic rather than a timer, and deliberately never auto-restarts a server it killed: a stopped server should stay stopped until you ask for it.
Production-grade async reverse proxy built on Axum and Tokio. JWT auth, rate limiting, IP filtering, dynamic config reload, per-route load balancing with health checks, Prometheus metrics export, and a Ratatui live TUI dashboard.
Ported core components of OpenCog's MOSES (Meta-Optimizing Semantic Evolutionary Search) genetic programming system to MeTTa, the language powering the next-generation Hyperon cognitive architecture. Involved deep symbolic AI, language semantics, and writing a test suite across two language runtimes.
Designed and facilitated 20+ internal training sessions on OpenCog concepts, MeTTa language fundamentals, and AGI theory for the iCog Labs team. Produced reusable training materials used onboarding new researchers.