I'm a researcher interested in how to build safe and reliable AI systems.
Currently my work focuses on reinforcement learning, large language models, and scaling relations in deep neural networks. Additionally my partner, Nova DasSarma, and I founded Hofvarpnir Studios, which is a non-profit dedicated to providing compute infrastructure and support to AI safety researchers.
My PhD is in theoretical physics, focusing on a few problems in the intersection of condensed matter and high energy theory. Topics of interest including: conformal field theories, dualities, non-Fermi liquids, topological states of matter, anomalies, etc. Prior to that I also dabbled in computational/quantum chemistry, doing QM/MM simulations with density functional theory.
I'm also an amateur mountaineer and rock climber, which has taken me to many beautiful places in the Sierras, Cascades, and beyond.