My name is Jiahao Fang, and I am currently a Master of Science (M.S.) student in Computer Science at the University of Illinois Urbana-Champaign (UIUC). My current research, supervised by Prof. Fan Lai, focuses on KV Cache Management for Large Language Model (LLM) Serving. I graduated from UIUC with a bachelor’s degree in Electrical Engineering and a minor in Computer Science.

During my undergraduate studies, I had the privilege of working as a research assistant with Prof. Daniel Kang on the AIDB project, which resulted in a publication at the DEEM workshop @ SIGMOD 2024. I also gained experience in machine learning and computational imaging through an independent study with Prof. Jane Zhao.

You can view my most recent CV, official transcripts from UIUC, official transcripts from ZJUI, and publications here.

Research Experience

• KV Cache Management for LLM Serving, University of Illinois Urbana-Champaign, Supervised by Prof. Fan Lai, August 2024 - Present
  1. Minimize Time to First Token (TTFT) for both Large Language Model (LLM) and Multimodal Large Language Model (MLLM).
  2. Overlap cache retrieval from multiple storage backends with the GPU-bound recomputing prefill.
  3. Working towards a publication, “Cross-layer KV Cache Parallelism for LLM Serving at Scale,” for OSDI 2026.
• The Application of Machine Learning to Database Query, University of Illinois Urbana-Champaign, Supervised by Prof. Daniel Kang, May 2023 - May 2024
  1. Contributed to the paper “AIDB: a Sparsely Materialized Database for Queries using Machine Learning,” accepted to the 8th DEEM workshop at SIGMOD 2024.
  2. Implemented and reproduced code for specify approximate selection with guarantees (SUPG).
  3. Generated datasets and ran experiments on query optimization for Applied AI for Database Systems and Applications (AIDB).
• Independent Study in Machine Learning and Computational Imaging, University of Illinois Urbana-Champaign, Supervised by Prof. Jane Zhao, January - May 2023
  1. Developed object detection models for unstructured datasets like images, audios, and videos.
  2. Summarized various Machine Learning algorithms in a Course Report based on mathematical proofs.