Zhaoyuan (Alex) Su
University of Virginia • acf7ea@virginia.edu
I am a fifth-year Ph.D. student in the DS2 Lab at the University of Virginia, where I am advised by Prof. Yue Cheng, working on LLM systems and systems for LLMs. I have also been working closely with Prof. Juncheng Yang at Harvard University on LLM serving systems.
My research focuses on building efficient and adaptive LLM systems, leveraging data-driven, reduction, and compression techniques to enhance performance, scalability, and reliability of LLM inference, storage, and deployment at scale.
Education
- University of Virginia, Ph.D. in Computer Science, GPA 4.0, Aug 2022 - Present
- George Mason University, Ph.D. in Computer Science, GPA 4.0, Aug 2021 - Jul 2022
Experience
- University of Virginia, Research/Teaching Assistant, Aug 2022 - Present
- Samsung Semiconductor, Research Intern, May 2024 - Aug 2024
- Argonne National Laboratory, Research Intern, May 2022 - Aug 2022
- George Mason University, Research/Teaching Assistant, Aug 2021 - May 2022
Publication
[arXiv Preprint]
MorphServe: Efficient and Workload-Aware LLM Serving via Runtime Layer Swapping and KV Cache Resizing.
[arXiv Preprint]
λScale: Enabling Fast Scaling for Serverless Large Language Model Inference.
[arXiv Preprint]
ZenFlow: Enabling Stall-Free Offloading Training via Asynchronous Updates.
[NSDI ’26]
Towards Efficient LLM Storage Reduction via Tensor Deduplication and Delta Compression.
(The 23rd USENIX Symposium on Networked Systems Design and Implementation (NSDI’26).)
[VLDB ’24]
Everything You Always Wanted to Know About Storage Compressibility of Pre-Trained ML Models but Were Afraid to Ask.
(50th International Conference on Very Large Data Bases (VLDB’24).)
[DRBSD ’22]
Understanding Impact of Lossy Compression on Derivative-related Metrics in Scientific Datasets.
(The 8th International Workshop on Data Analysis and Reduction for Big Scientific Data (DRBSD’22), affiliated with SC’22.)
Social Links
- Github: https://github.com/alexsssu
- LinkedIn: https://www.linkedin.com/in/zhaoyuan-su-b40845192/
- Website: https://alexsssu.github.io/