I am an AI safety researcher at Koç University, working on the gap between what we ask AI systems to do and what they actually optimize for. My research spans both empirical and theoretical directions — from cataloging specification failures in the wild to developing RL algorithms with formal pessimism guarantees.