I am Yusen Peng, an first-year CSE PhD student at The Ohio State University (OSU), starting in Fall 2026. I am incredibly fortunate to be advised by Dr. Sachin Kumar. My current research interest is efficient multimodal LLMs - How can we compress tokens and/or model parameters in a principled way for multimodal LLMs in order to improve both training and inference efficiency?
I earned my B.S. in Computer Science and Engineering from The Ohio State University with Summa Cum Laude and Honor Research Distinction. Previously, I worked on a variety of AI/ML topics including time series analysis (pytskit, TSB-FCST), 3D vision (SVD^3), action recognition (CascadeFormer), and mechanistic interpretability (CE-Bench).