Building reliable software systems where application engineering meets AI.
I design and build production-oriented software across agent runtimes, developer tools, web platforms, and distributed services. My work focuses on making complex systems observable, testable, and safe to operate.
| Area | Focus |
|---|---|
| Agent systems | Tool calling, runtime loops, policy checks, event streams, checkpoint/resume, human-in-the-loop |
| AI & knowledge | RAG, hybrid retrieval, reranking, evaluation datasets, model-provider abstractions |
| Backend & platform | Python, FastAPI, Node.js, Go, PostgreSQL, Redis, Docker |
| Frontend & tooling | TypeScript, React, Webpack, browser extensions, plugin ecosystems |
| Reliability | Automated testing, schema validation, observability, metrics, tracing, security boundaries |
- agent-runtime-examples — A reference Agent Runtime with durable events, SSE replay, tool policies, workspace controls, verification, leases, and optional distributed infrastructure.
- Make state transitions explicit.
- Treat events as a durable source of truth.
- Keep policy checks before side effects.
- Prefer deterministic tests and reproducible examples.
- Document trade-offs, limitations, and operational boundaries.
Python · TypeScript · JavaScript · Go · FastAPI · React · PostgreSQL · Redis · Docker · Prometheus · Grafana
Selected projects and engineering notes are maintained in the repositories below.