About
I am an AI Safety researcher based at LISA in London, leading the Heuristic Arguments research group at Iliad. This includes mentoring Iliad fellows, MATS fellows and teaching at the Iliad Intensive. I’m hiring! If you would like to work with me please send me an email with your CV, and I’ll be in touch.
From June 2024 to May 2026 I was a researcher for the Alignment Research Center. Prior to that, I did a PhD from 2019-2024 at the University of Oxford under the supervision of Prof Sir Andrew Wiles. My main mathematical interest is number theory - specifically Iwasawa theory, the Langlands program and p-adic L-functions. My thesis was on equidistribution problems over totally real fields.
I did my undergraduate and masters degrees in Mathematics at Cambridge University, where I was a member of Gonville & Caius College.
AI Alignment
Estimating the expected output of wide random MLPs more efficiently than sampling - joint with Wilson Wu, Victor Lecomte, Michael Winer, Jacob Hilton and Paul Christiano
Analytically Learning Variational Auto-Encoders - joint with David Matolcsi and Paul Christiano
Mathematics
Here is my My PhD Thesis, submitted in May 2024.
Some talks I’ve given:
An introductory talk to a seminar on the Langlands Program
A talk on Galois representations coming from automorphic representations