I’m Uzay (pronounced ooze-eye). I work on scaling human oversight and understanding with AI at Fulcrum Research.

At MIT, I studied many pretty ideas. Some of them are quantum mechanics, complexity theory, abstract algebra, inference and information, and statistical mechanics.

In the past, I’ve done research on the dynamics of in context learning, synthetic generation of challenging software tasks, scaling laws for language modeling, and interpreting LLM adversaries. I also did a lot of coding, and made a bunch of open source software.

I grew up in Paris, and then lived in pika around MIT. I like art. I write poetry and essays on my blog and my substack. I’m an avid climber and I would like to live in the mountains many years from now.

Email: zef (at) mit (dot) edu

twitter, github, goodreads