Professor of Computer Science, Princeton University
Director of Princeton AI Alignment and Safety (AIAS)
Co-founder, Google AI Princeton
I study the foundations of machine learning, optimization, and control, developing algorithms that learn from experience, decide under uncertainty, and predict and control dynamical systems. I also study how online learning, game theory, and mechanism design can help evaluate and align increasingly capable AI.
See research, bio and CV, and our group webpage and blog; visit Princeton AIAS for alignment research, seminars, and opportunities.
Selected recent research
- Measuring Intelligence Beyond Human Scale (SepaRank, 2026): evaluating models with challenges that distinguish other models.
- AI Alignment via Incentives and Correction (2026): designing rewards for effective solver–auditor interaction.
- Universal Sequence Preconditioning (2025): improving sequential prediction with polynomial preconditioning.
All publications on Google Scholar · Most recent first · Research overview
Books and surveys
- Introduction to Online Convex Optimization
- Optimization for Machine Learning
- Introduction to Online Nonstochastic Control
Selected talks
- The Economics of AI Alignment — Part 1, Princeton AIAS, October 9, 2026 · Video
- Provably Efficient Learning in Nonlinear Dynamical Systems via Spectral Transformers, MIT LIDS, March 10, 2026 · Video
- Optimization for Machine Learning
- The Nonstochastic Control Problem
More talks and lectures on my YouTube channel.
Teaching
Courses: Theoretical Machine Learning, AI Alignment and Safety, Computational Control Theory, and Introduction to Machine Learning. Course websites, semesters, and tutorials.