Rajan Agarwal

Multimodal retrieval and video search researcher — Rajan focuses on treating video as structured, four‑dimensional data and building retrieval/indexing systems that find "needles" (specific moments) in hours of footage. He has presented this work publicly (Kino AI talk) and iterated on practical systems for filmmakers and editors. youtube+1

Agent & RL practitioner — he builds and experiments with interoperable AI agents (Human Simulation Lab) and coding agents (Shadow), exploring alignment, tool usage, and multi-agent orchestration to perform complex end-to-end tasks (e.g., prototype startup built entirely by agents). humansimulation+1

Hands-on systems and robotics engineer — active in student autonomous-vehicle projects (WATonomous), practical low-latency controllers, and hardware-in-the-loop work; blends systems engineering with ML to ship working demos and compete at hackathons. rajan+1

Open-source & student leadership — maintains many public repos (ML, tools, notebooks), participates in hackathons and student teams at Waterloo, and has founded/led small research and product efforts (Shapeshift, Human Simulation Lab). github+2