Ph.D. student · Computer Science
University of Illinois Chicago
I'm a Ph.D. student at UIC, advised by Prof. Philip S. Yu in the Big Data and Social Computing Lab.
Previously, I earned my B.S. and M.S. in Computer Science at UC San Diego, where I worked with Prof. Zhiting Hu.
Exploring how AI systems learn, reason, and interact with people and the world.
Sycophancy Suppression Can Impair Rational Updating is now on arXiv, accepted to EMNLP 2026 Findings. We study how to reduce sycophancy while preserving a model’s ability to learn from useful feedback.
Read more about this August 2026 updateIntroducing MemAudit, our benchmark for measuring what AI agents actually remember about their users across interactions.
Read more about this June 2026 updateOur survey, The Landscape of Agentic Time Series Systems, is available, covering system architectures, reliability, and open research directions.
Read more about this June 2026 updateARC-STAR is on arXiv. We develop auditable corrections for PDE foundation models, focusing computation on regions that need refinement.
Read more about this May 2026 updateLocally Confident, Globally Stuck is on arXiv, studying quality versus exploration in diffusion language models. Update: accepted to COLM 2026 in July.
Read more about this April 2026 update02 / Research
My research spans AI agents, large language models, and multimodal learning, with the goal of building more capable and adaptable AI systems. I'm interested in how AI systems can better understand the world and apply what they learn to perform tasks more effectively.
Memory, reusable skills, and learning from experience to help agents generalize across tasks and environments.
Understanding and improving how language models reason, explore solutions, and respond to new evidence.
Connecting language, vision, and other modalities for richer understanding of the world, with an interest in world models.
03 / Selected work
Findings of EMNLP 2026 · Accepted
Evaluating how anti-sycophancy interventions affect a model's ability to revise answers when new evidence is valid.
arXiv 2026 · Preprint
A benchmark that audits what an agent's long-term memory actually retains about its users.
COLM 2026 · Accepted
Understanding and improving the trade-off between answer quality and reasoning-path exploration in diffusion LLMs.
NeurIPS 2025 · Poster
LLMs reason directly over single-cell data and bioinformatics tools to produce interpretable scientific analyses.
04 / The journey
Ph.D. in Computer Science
Advisor: Prof. Philip S. Yu · BDSC LabResearch Intern
Advisor: Prof. Zhiting HuM.S. in Computer Science
B.S. in Computer Science
Let's connect
Always happy to talk about research, collaborations, or an interesting idea.