I'm a forward deployed engineer in practice: I take systems into enterprise customer environments and get them working in production. Right now that means warehouse robotics at Geek+, where I deploy and debug robot fleet orchestration (RMS), WMS integrations, SSO and infrastructure at client distribution centers.
These days I'm building LLM and agent tooling: MCP servers, retrieval with citations, and privacy tooling for enterprise AI.
- 🛡️ Building canpii, Canadian PII recognizers for Microsoft Presidio
- 🔭 Building rmf-mcp, an MCP server that lets LLM agents operate robot fleets through Open-RMF
- 🌱 Contributing to Open-RMF and FastMCP
- 💼 Background: finance and risk (Columbia MS Enterprise Risk Management, UBC Sauder BCom Finance), then software engineering, then on-site deployment
- 🌐 English / 中文
- 📫 LinkedIn · Toronto, Canada
| Project | What it does | Stack |
|---|---|---|
| canpii | Canadian PII recognizers for Presidio: provincial health numbers, passports, CRA business numbers, bank and driver's licences. Uses English/French context, exports to LiteLLM and MCP gateways, and ships PIPEDA / PHIPA / Quebec Law 25 presets. | Python, Presidio |
| rmf-mcp (in progress) | Exposes Open-RMF fleet state and tasks to LLM agents over MCP, so an agent can query robots and dispatch tasks in simulation. | Python, MCP, Open-RMF, ROS 2 |
| KEEPER (internal, not public) | Internal knowledge base that turns Q&A from team meetings into cited, searchable entries; available to engineers through an enterprise-chat bot. | LLM pipeline, RAG, WeCom bot |
- Docketeer (OSLabs): Docker and Kubernetes developer tool for managing containers and visualizing cluster and container metrics
Languages: Python · TypeScript · SQL · Bash AI: RAG · MCP · agents · evals Infra: Linux (RHEL) · nginx · Docker · Kubernetes · MySQL · Redis · GCP · Azure

