|
Fangzhou Wu I'm a third-year Ph.D. student at University of Wisconsin–Madison, where I am fortunate to be co-advised by Kristin Eschenfelder and Sandeep Silwal. Prior to coming to Madison, I earned my bachelor's degree from Huazhong University of Science and Technology (HUST). |
|
ResearchI am broadly interested in developing provably efficient algorithms to accelerate training and inference for foundation models and agents. My research aims to bridge theoretical insights and practical system design by integrating these algorithms into modern foundation-model-based applications. |
Selected Work |
|
Randomization Boosts KV Caching, Learning Balances Query Load: A Joint Perspective
Fangzhou Wu, Sandeep Silwal, Qiuyi (Richard) Zhang ICLR, 2026 code / arXiv |
|
Efficient Training-Free Online Routing for High-Volume Multi-LLM Serving
Fangzhou Wu, Sandeep Silwal NeurIPS, 2025 code / arXiv |
|
System-Level Defense against Indirect Prompt Injection Attacks: An Information Flow Control Perspective
Fangzhou Wu, Ethan Cecchetti, Chaowei Xiao arXiv, 2024 code / arXiv |
|
A New Era in LLM Security: Exploring Security Concerns in Real-World LLM-based Systems
Fangzhou Wu, Ning Zhang, Somesh Jha, Patrick McDaniel, Chaowei Xiao arXiv, 2024 project page / arXiv |
Academic Service |
|
| Reviewer, ICLR | 2024–2026 |
| Reviewer, NeurIPS | 2024–2025 |
| Reviewer, ICML | 2024, 2026 |
Teaching |
|
| TA, Applied Database Design (LIS 464), UW–Madison | SP24–FA25 |