I am a Ph.D. student in Computer Science at the University of Southern California, advised by Prof. Yue Zhao in the FORTIS Lab.

I study how to understand and improve AI, and how it shapes science and society.

LLM Evaluation and Post-Training. I view evaluation as a way to discover what models can do and what they should learn next. I develop benchmarks and automated probing methods to explore the frontiers of model capabilities, seeking challenges that reveal both their current limits and opportunities for growth. Through post-training, I aim to translate these insights into new and more generalizable capabilities.

AI, Science, and Society. I am interested in how increasingly capable AI can expand what humans are able to understand and accomplish. This requires more than producing useful outputs: it requires making sense of AI-generated ideas and results through interpretability and other approaches, examining their validity, and building upon them to create knowledge and guide action. I explore these questions through AI for scientific discovery and real-world problem solving, with the broader aim of extending our ability to investigate, understand, and change the world.

Previously at WeChat AI, Alibaba, MBZUAI, the MINE Lab at Notre Dame, MIT CTL, Microsoft, and Sichuan University.