CS PhD candidate at Purdue University, advised by Prof. Xiangyu Zhang.
Background
Research interests include: LLM Red-teaming, RL for Security, LLM Agents Evaluation, Backdoor Attacks and Defenses. Specifically, he studies attacks and defenses for LLMs, LLM agents, and RL systems, including jailbreaking; backdoor attacks, detection, and removal; and broader adversarial robustness. He combines reinforcement learning with programmatic test generation to discover vulnerabilities and failure modes and to develop principled defenses that harden models and agents. Recently, he is also working on building security-focused LLM agents and stress-testing of LLM agents across key evaluation dimensions.