Yi Xu
 |
Research Scientist, |
About me
I'm a Research Scientist at Meta. Previously, I was a Postdoc at Sky Computing Lab, UC Berkeley, working with Prof. Ion Stoica. I earned my PhD from UC San Diego, under the supervision of Prof. Steven Swanson. My main research interests span memory and storage management, programming models, distributed systems, and their intersections. Recently, I've been focusing on GPU and LLM memory management.
Selected Publications
Crossflow: Prefill-Decode Elasticity for Agentic LLM Serving
Yi Xu, Ehsan K. Ardestani, Wenyin Fu, Martin Schatz, Krishna Malladi, Zhan Shu, Adnan Aziz, Shobhit Kanaujia, Ajit Mathews, Chunqiang Tang
[arXiv preprint 26] [paper]EMA: Elastic and Performance Transparent Memory Across GPUs
Yi Xu, Tian Xia, Ion Stoica
[arXiv preprint 26] [paper]Pie: Pooling CPU Memory for LLM Inference
Yi Xu, Ziming Mao, Xiangxi Mo, Shu Liu, Ion Stoica
[arXiv preprint 24] [paper]CXL Shared Memory Programming: Barely Distributed and Almost Persistent
Yi Xu, Suyash Mahar, Ziheng Liu, Mingyao Shen, Steven Swanson
[arXiv preprint 24] [paper]IONIA: High-Performance Replication for Modern Disk-based KV Stores
Yi Xu*, Henry Zhu*, Prashant Pandey, Alex Conway, Rob Johnson, Ramnatthan Alagappan, Aishwarya Ganesan (*co-first authors)
[FAST 24] [paper]Zhuque: Failure is Not an Option, it’s an Exception
George Hodgkins*, Yi Xu*, Steven Swanson, Joseph Izraelevitz (*co-first authors)
[ATC 23] [paper]Clobber-NVM: Log Less, Re-execute More
Yi Xu, Joseph Izraelevitz, Steven Swanson
[ASPLOS 21] [extended abstract] [paper] [code]Basic performance measurements of the intel optane DC persistent memory module
Joseph Izraelevitz, Jian Yang, Lu Zhang, Juno Kim, Xiao Liu, Amirsaman Memaripour, Yun Joon Soh, Zixuan Wang, Yi Xu, Subramanya R Dulloor, Jishen Zhao, Steven Swanson
[arXiv preprint 19] [paper]
Other Publications
Breaking the Tradeoff: Elastic and Isolated GPU Sharing with Ghost
Yicheng Liu*, Yifan Qiao*, Tian Xia, Yi Xu, Tony Hong, Yutong Zhou, David Sun, Shuo Yang, Yilong Zhao, Junyi Shu, Jiarong Xing, Harry Xu, Ion Stoica, Joseph Gonzalez, Sam Kumar (*co-first authors)
[ASPLOS 27, to appear]SkyNomad: On Using Multi-Region Spot Instances to Minimize AI Batch Job Cost
Zhifei Li*, Tian Xia*, Ziming Mao, Zihan Zhou, Ethan J. Jackson, Jamison Kerney, Zhanghao Wu, Pratik Mishra, Yi Xu, Yifan Qiao, Scott Shenker, Ion Stoica (*co-first authors)
[NSDI 27, to appear] [paper]Idleness is Relative: Exploiting Tool-Call Idle Windows for Offloading in Agentic Systems with MORI
Tian Xia, Hanchen Li, Zhifei Li, Xiaokun Chen, Hao Kang, Yifan Qiao, Yi Xu, Ion Stoica
[arXiv preprint 26] [paper]
Workshops
Zhuque: Failure is Not an Option, it’s an Exception
George Hodgkins*, Yi Xu*, Steven Swanson, Joseph Izraelevitz (*co-first authors)
[NVMW 23] [paper] [poster] [slides] [video]Clobber-NVM: Log Less, Re-execute More
Yi Xu, Joseph Izraelevitz, Steven Swanson
[NVMW 21] [paper] [slides]
Education
| Sept. 2018 – Aug. 2023 | Ph.D. in Computer Science and Engineering, UC San Diego, USA |
| Sept. 2014 – June 2018 | B.Sc. in Computer Science, Southeast University, China |
Experience
| Aug. 2025 – Present | Research Scientist, Meta |
| Aug. 2023 – Aug. 2025 | Postdoc, Sky Computing Lab, UC Berkeley (advisor: Prof. Ion Stoica) |
| June 2022 – Sept. 2022 | Research internship, VMware Research Group |
| June 2021 – Sept. 2021 | Research internship, VMware Research Group |
| June 2019 – Sept. 2019 | Research internship, Oracle Inc |
| July 2017 – Sept. 2017 | Research internship, The University of Hong Kong (HKU) |