I am a Research Scientist at Meta SuperIntelligence Labs (MSL), working on scaling post-training and inference for foundation models. I co-created MSL’s research inference stack for frontier AI model research. Previously, I worked on high performance AI infrastructure networking at Meta.
I obtained my Ph.D. from the University of Washington, where I was fortunate to work with Ratul Mahajan and Arvind Krishnamurthy on network verification. Prior to that, I got my B.S. from Peking University, where Chenren Xu introduced me to computer networks research.
Current Work
At MSL, I work on the research infrastructure that enables frontier AI model research at scale. My focus spans two complementary areas:
LLM Inference for Research: Building inference systems tailored to the needs of research — where models change rapidly and correctness against the training stack is critical.
Post-Training Scaling: Scaling RL training for foundation models. Solving challenges from both the system side and ML side.
Due to the nature of this work, most details are confidential.
PhD Research
My PhD dissertation addresses reliability risks in network infrastructure operation. It pioneers coverage metrics for network testing [1] [2] and develops a formal language to describe network changes concisely and verifiably [3]. This work enabled engineers to precisely specify and rigorously verify network operations before rollout — a new paradigm for a field historically dominated by caution and manual processes.
A significant portion of this research was built into production systems and deployed in some of the world’s largest networks [1] [3]. My work [2] received an Applied Networking Research Prize from IRTF (Internet Research Task Force) [4].
Publications
Relational Network Verification. In SIGCOMM 2024
Xieyang Xu, Yifei Yuan, Zachary Kincaid, Arvind Krishnamurthy, Ratul Mahajan, David Walker, Ennan Zhai.
[PDF] [Code]
Test Coverage for Network Configurations. In NSDI 2023
Xieyang Xu, Weixin Deng, Ryan Beckett, Ratul Mahajan, David Walker.
[PDF] [Code] [Demo] [Talk]
Test coverage metrics for the network. In SIGCOMM 2021
Xieyang Xu, Ryan Beckett, Karthick Jayaraman, Ratul Mahajan, David Walker.
[PDF] [Talk]
PassiveVLC: Enabling practical visible light backscatter communication for battery-free IoT applications. In MobiCom 2017
Xieyang Xu, Yang Shen, Junrui Yang, Chenre Xu, Guobin Shen, Guojun Chen, Yunzhe Ni.
[PDF] [Intro Video]
Contact
xieyang.uw at gmail dot com
Last updated: Apr 2026