news

Oct 2026 Joined Agents’ Last Exam v2 at Berkeley AI Research as a contributor on the science track.
July 2026 ContextLeak, our black-box privacy auditing framework for private in-context learning, was accepted to COLM 2026.
June 2026 Started as a Student Research Assistant and Visiting Student at Stanford University.
2025 Our work on auditing privacy-preserved in-context learning methods was accepted to the ACL 2025 L2M2 Workshop.