About Me
I am Kaiwen Shi, a Ph.D. student at the University of Notre Dame. My research explores how language model agents can reason with structured knowledge, collaborate effectively, and express uncertainty more faithfully.
Previously, I studied Computer Science with a minor in Economics at Yuanpei College, Peking University.
News
- Our paper SAGE has been accepted to EMNLP 2026!
- Our paper Drift-Bench has been accepted to ICML 2026!
- Our paper AgentRouter has been accepted to ACL 2026 Main!
- Our paper NG-Router has been accepted to EACL 2026 Main!
- Our paper BLURR has been accepted to the WWW 2026 Demo Track!
Education
- University of Notre Dame — Ph.D. student.
- Peking University, Yuanpei College — Undergraduate degree in Computer Science, with a minor in Economics.
Research interests
- Multi-agent collaboration: task-aware routing and adaptive coordination for question answering.
- Uncertainty and alignment: uncertainty quantification, confidence calibration, and policy optimization.
- Knowledge-grounded reasoning: knowledge graphs, retrieval, and multi-hop reasoning.
- Trustworthy AI: citation verification, robust agent evaluation, and health and social-service applications.
Publications
* Equal contribution