Junze Ye (Tony)
I work on efficient post-training and alignment of AI models under data or compute constraints. My recent work includes on-policy teacher data collection for LLM agents, adaptive physician querying for clinical benchmark labeling, and inference-time fusion of generative experts.
I am a 4-th year PhD candidate at Stanford University. I am fortunate to be advised by Prof. Mohsen Bayati and supported by the Program in Operations, Information, and Technology. Previously, I completed my B.S.E. at Princeton University, where I worked with Prof. Jason Klusowski on statistical machine learning and model interpretability.