Yao Fu (符 尧)
Ph.D. Student at The University of Edinburgh

1.45, Informatics Forum
Edinburgh, EH8 9AB
Scotland, UK
I am a Ph.D. student in Computer Science at The University of Edinburgh, supervised by Prof. Luo Mai. I received my B.Eng. degree in Computer Science and Technology from Sun Yat-sen University in June 2021. I was supervised by Prof. Di Wu at Sun Yat-sen University as a member of Yat-sen Honor School.
I study the intersection of machine learning and distributed systems. My goal is to build efficient systems for the large-scale deployment of machine learning models. My current research focuses on the efficient inference of large language models in serverless computing clusters.
news
Jan 22, 2025 | I’ll be visiting University of Pennsylvania (29 Jan) and Columbia University to present our latest work on Serverless AI, focusing on the efficient sharing of AI infrastructures. Check details here. I’ll be in Philly, DC, and NYC from Jan 27 to Feb 5. If you’re in the area and interested in having a chat, let’s meet up! |
---|---|
Oct 01, 2024 | I’m honored to serve as a reviewer for IEEE Transactions on Mobile Computing (TMC). I’m excited about the opportunity to contribute to the community in this new role! |
May 16, 2024 | I’m selected as one of the ML and Systems Rising Stars! Thanks to everyone who has supported me along the way! I’ll be attending the workshop at NVIDIA’s headquarters in Santa Clara, CA, on July 15-16. |
Mar 21, 2024 | Our paper “ServerlessLLM: Locality-Enhanced Serverless Inference for Large Language Models” has been accepted to OSDI 2024. Preprint available on ArXiv. Code will be released soon. Stay tuned! ![]() ![]() |
Jan 25, 2024 | We released two new papers on ArXiv! Check them out: ServerlessLLM and MoE-Infinite |
publications
- ServerlessLLM: Locality-Enhanced Serverless Inference for Large Language ModelsOSDI, 2024
- MoE-Infinity: Activation-Aware Expert Offloading for Efficient MoE ServingarXiv preprint arXiv:2401.14361, 2024
- TorchOpt: An Efficient Library for Differentiable OptimizationJMLR, 2023
- Ekko: A Large-Scale deep learning recommender system with Low-Latency model updateOSDI, 2022