Peter W. Hoffman

I’m a 2nd-year PhD student at MIT LIDS working on generative artificial intelligence with Professor Pablo Parrilo and Professor Gabriele Farina. I am mainly interested in problems at the intersection of deep learning, probability, and alignment, particularly those related to diffusion and human models.

Previously, I completed my undergraduate degree in Mathematics, also at MIT, where I worked with Professor Pablo Parrilo on optimization algorithms with applications in game theory. I am also involved with teaching: in the past, I’ve helped design and teach 6.S095, a student-run lecture series at MIT covering advanced problem-solving techniques in probability, random variables, and stochastic processes. I also frequently post notes on a variety of topics, which you can find here.

In the summer of 2026 I was an Anthropic-MIT Fellow based out of San Francisco where I worked with the Applied AI and Beneficial Deployments team in collaboration with MIT on alignment and evaluation-related problems. I’ve also spent time as a quantitative researcher at Point72 and Citibank.

If I’m not writing math or code, you can probably find me painting. Feel free to reach out at hoffmanp@mit.edu.