I am a Senior Researcher in the Intelligent Networked Systems Group at Microsoft Research Redmond. In the spring of 2025, I will be joining the University of Illinois at Urbana-Champaign (UIUC) as an Assistant Professor of Computer Science, with an affiliate appointment in Electrical & Computer Engineering.

My research builds and optimizes networked systems with practical machine learning (ML), ranging from developing large-scale systems and research platforms that lay the foundation for applying ML, to devising practical ML algorithms with mechanisms that improve their real-world deployability in systems and networks.

ML-enhanced video systems are a major thrust of my work. I am the creator of Puffer, a live TV service with over 360,000 real users, and a video research platform that has aided researchers in publishing top-tier and award-winning papers.

My vision is to enable a future of self-evolving systems, which intelligently and autonomously adapt their design (architectures and algorithms) and implementation (code and configuration) in response to changing environments and user behaviors. The emergence of large language models (LLMs) has allowed us to take a first step toward this vision.

Prior to Microsoft Research, I received my Ph.D. in computer science from Stanford University, where I was advised by Keith Winstein and Philip Levis. Before that, I did my undergraduate studies at Tsinghua University (Yao Class) and MIT. I am a recipient of the IRTF Applied Networking Research Prize, the USENIX NSDI Community Award, the USENIX NSDI Outstanding Paper Award, the USENIX ATC Best Paper Award, and the APNet Best Paper Award.

ML-enhanced video systems

• Puffer [NSDI ’20]
• Tambur [NSDI ’23]
• Grace [NSDI ’24]
• OpenNetLab [APNet ’22]

ML for networking

• Pantheon [ATC ’18]
• Genet [SIGCOMM ’22]
• Teal [SIGCOMM ’23]
• Diffy [PLDI ’24]

ML-assisted cloud applications

• Autothrottle [NSDI ’24]
• (more to come)

Self-evolving systems (LLM-based)

• Nada [HotNets ’24]
• (more to come)

Please visit my research page for more details.