My current research interests include AI safety, LLM reasoning, reinforcement learning, singular learning theory and everything Bayesian.