Projects

A coordination product where your AI assistant negotiates meetings with other people's assistants. Assistants connect over MCP (plus REST and A2A), scheduling works from free/busy only, and anything that commits you waits for your approval.

This site: an agentic personal site. Instead of a static portfolio, a chat agent answers questions about my work from markdown content, checks my availability, and books time with me. React and Vite on Vercel; Python, FastAPI, and Google ADK with Gemini on Render.

Project lead for Tesla Factory Software's AI agent platform: a shared agent runtime, auth and session services, a React chat client, observability, and multi-site Kubernetes deployment, plus a portfolio of production agents used daily by supervisors, engineers, technicians, operators, and quality and inventory teams.

Plain-English answers for factory floor workers over 1M+ pages of equipment manuals and procedures, grounded in and citing the source documents. Hybrid dense and BM25 retrieval on PostgreSQL with pgvector, served by a Go microservice on Kubernetes.

Onboarded 300+ engineers across 10+ teams to AI-native development, and built a SWE Agent with access to Kubernetes, Splunk, Grafana, GitHub, Confluence, and Jira that triages issues, plans sprints, and reports status in place of standups and sprint planning.

Lead engineer for the factory software suite (CMMS and spare-parts management) tracking 120K assets and 2M parts, with PLC telemetry from 5,000+ robots feeding predictive-maintenance analytics on an event-driven Go, gRPC, and Kafka service mesh.

One invoicing platform for Service, Supercharging, and e-commerce, processing $4B+ a year at 99.99% accuracy. Declarative billing rules and a Python tax engine cut billing latency from days to minutes.

Tesla's in-house HR system: performance reviews, versatility charts, onboarding, and Avature ATS integration, scaling from 30K to 130K employees.