Ziyang Li

AI student at Zhejiang University · LLM interpretability & systems efficiency

Junior in Artificial Intelligence at Chu Kochen Honors College (GPA 3.93/4.00). My work sits at the intersection of ML systems and model interpretability — currently focused on agent efficiency, and previously on understanding the causal structure of intermediate Transformer layers.

ThunderAgent: A Simple, Fast and Program-Aware Agentic Inference System

Hao Kang*, Ziyang Li*, Xinyu Yang*, Weili Xu*, Yinfang Chen, Junxiong Wang, Beidi Chen, Tushar Krishna, Chenfeng Xu, and Simran Arora

Preprint, Feb 2026