I am a Ph.D. student in Computer Science at the University of Illinois Urbana-Champaign, supported by a CS Ph.D. Fellowship and advised by Prof. Yiling Lou. I work on reliable long-horizon coding agents and agentic AI: when a partial agent trajectory already reveals its eventual outcome, when an agent should stop or ask for help, and how to evaluate either without the leakage and contamination that quietly invalidate agent benchmarks.

Previously, I was a Research Affiliate at the MIT Media Lab, working with Prof. Paul Liang, and a Research Assistant at Johns Hopkins University, collaborating with Prof. Hao Frank Yang and Prof. Hongru Du. That work spanned multimodal large language models and human-centered AI, including a NeurIPS 2025 Spotlight paper on personalized decision modeling.

Before academia, I worked as a Software Engineer at Microsoft, where I built and operated large-scale distributed services in Azure Workflow — production infrastructure, incident response, and an early LLM-over-logs diagnostics tool that is where my interest in agent observability started.

I received my B.S. in Computer Science from Tongji University, where I worked on medical imaging and robotics with Prof. Yufei Chen and Prof. Peng Qi.