News

Our work, Weight space gradient descent induces approximate activity descent, making learning rules testable, has been accepted to the NeurReps workshop at NeurIPS 2026 in Sydney!

My new preprint, Improving scalable oversight with co-trained monitors, studies how co-training monitors alongside workers can make AI oversight more robust to monitor evasion.

I received the 2026 Donald S. Murray Prize for Excellence in Teaching from the Statistics and Data Science Department at the Wharton School, University of Pennsylvania.

Our paper, Estimating Implicit Regularization in Deep Learning, with Kevin Tan, Giles Hooker, and Konrad Kording, has been accepted to NeurIPS 2026 in Sydney!

Received $2,500 in funding from the Wharton AI & Analytics Initiative for the proposal, Teaching statistics and data science from first principles; a data-oriented, AI-enabled course, with Adi Wyner.

Spending a semester at Tel Aviv University as a visiting scholar in the Blavatnik School of Computer Science and AI, hosted by Nadav Cohen. Sponsored by TAU’s Research Fellowship for Outstanding Students.

Attended the 2025 Summer School on Analytical Connectionism hosted at the Gatsby Computational Neuroscience Unit at University College London.