Software engineering student in Montréal, working on applied AI — retrieval, computer vision, and LLM-backed tools — with a bias toward systems that run and get measured rather than notebooks that demo well.
Clypio is where most of my time goes. It's a performance clipping network: creators cut a customer's long-form video into shorts and post them, and the customer is billed per view that survives verification rather than per view reported. I build and run the verification engine behind it — view ingestion from platform reporting, bot screening, per-clip caps, and the billing ledger that falls out of those. YouTube Shorts verification is live and TikTok is in platform review. The engine is private; the site is at clypio.com.
Other things worth reading:
- ai-repo-inspector — a repository review tool with CLI and MCP front ends. Probably the best window into how I work: the writeup covers an arbitrary file write I found in ref handling and an MCP tool that returned a confidently wrong answer on every call without ever erroring.
- document-assistant — local RAG over your own documents and code. sentence-transformers for embeddings, Chroma for storage, Ollama for generation, so nothing leaves the machine.
- meetingmind — an Electron prototype that records a meeting, transcribes it, and drafts the follow-up. Prototype-grade and labelled as such.
Currently studying Software Engineering at Concordia. Open to internships, junior roles, and contract work. English, French, and Arabic.