Harvard University — Student
Studying and doing research in computer vision and video models.
A quick note
This is a minimalist, text-first page. Links behave like citations: obvious, consistent, and easy to scan.
I care about models that work on real video: long horizons, fast inference, and evaluation that reflects reality. When I’m not thinking about video, I’m probably thinking about food.
What I work on
Topics I keep coming back to, with an emphasis on video models and real-time constraints.
Representation learning and generation for video, with attention to temporal coherence and evaluation that matters.
Fast inference, streaming settings, and systems constraints that shape model design.
Learning policies that act in the world, and thinking about how to evaluate and deploy them responsibly.
Selected experience
A short list. If you want a full CV, email me.
ByteDance Seed — Research / Engineering
Worked on research and systems around modern vision and video modeling.
Decart — Research / Engineering
Explored video and real-time constraints with an engineering mindset.
Notes
For now: this section is intentionally small. I may add posts and reading notes later.
If you’re looking for a specific topic, email me and I’ll point you to relevant work.
Currently enjoying: video transformers, long-video benchmarks, and anything that makes evaluation less hand-wavy.
Best way to reach me
Email works. If you’re sending a cold email, a single paragraph with context is perfect.