Software Engineering student at Carleton University. I build AI products end to end: mobile and web apps, backend jobs, retrieval, and tools for AI agents. Open to a Winter 2027 co-op.
- 465Cal · AI nutrition app on iOS, Android and web. Meal photo → vision model → USDA lookup (~13,590 records, hybrid keyword/vector search) → calories and macros, with Gemini → GPT-4o-mini failover. The code is private; the 500-dish Nutrition5k benchmark is public, limitations included.
- Ascended Browser · hardened browser MCP server for Claude Code, Codex and opencode. On PyPI and npm.
- Odysseus Web MCP · web search and fetch MCP server with SSRF protections, listed in the official MCP Registry. Passed a live 20-case qualification run.
- VecterDB-Q · semantic search engine with Flat and IVF indexes written from scratch, FastAPI, Docker, 41 tests.
- RAG-chunker · turns PDFs into RAG-ready chunks with embeddings and BM25.
- 1st place, SKYTRAC 2026 Hackathon: push-to-talk app for remote teams over satellite networks. I wrote most of the backend.
- 3rd place twice in the 2026 Red Team DefTech Hackathon Series: Ottawa (cryptographic supply-chain provenance, CanadaChain) and Toronto (autonomous agent for a comms-denied vehicle simulator).
- Backend intern at Dribbl (2025): +42% search relevance with domain-specific embeddings, 800 ms → 200 ms with vector caching.
- Built HiringMessage (2025–2026), an AI resume-tailoring tool with paying users.
- Contributed CI/CD test infrastructure to trycua/cua.