Yunming Xiao
肖蕴明
Assistant Professor @ CUHK-Shenzhen
国家级青年人才
Room 403a, Zhi Xin Building
2001 Longxiang Road
Longgang District, Shenzhen, China
About Me
Hi there! I am an Assistant Professor at the School of Data Science at The Chinese University of Hong Kong (CUHK), Shenzhen. Before joining CUHK Shenzhen, I was a Research Fellow at the University of Michigan working with Prof. Ang Chen. I received my Ph.D. in Computer Science from Northwestern University in 2024, advised by Prof. Aleksandar Kuzmanovic, and received my B.Eng. in Computer Science from Beijing University of Posts and Telecommunications in 2019.
I am broadly interested in computer systems, networks, and security. My work has received the Best Paper Award from APNet and the Best Student Paper Award from ACM EuroSys. My current research focuses on the following topics:
-
[⭐New!] Agent-native Web infrastructure. AI agents are reshaping the online ecosystem, creating a need for infrastructure designed around how they retrieve information, reason, and act. Our Semantics Delivery Network (SemDN) delivers relevant passages instead of whole pages, reducing repeated fetching and processing while improving retrieval quality per context token. This vision opens questions in semantic search, caching, freshness, and security. We are also exploring the Internet of Agents.
-
Secure Internet and AI services. We ask how users and content providers can benefit from Web intermediaries without fully trusting them. A line of work focuses on CDN, spanning private edge analytics [Snatch, TOCS'25], secure password pre-authentication [PreAcher], and encrypted WAFs [PhantomBox]. This agenda extends to distributed Internet services [RING, Horizon], private DNS [PDNS], DDoS defense [Canopy], online advertising [Obsidian], and increasingly AI services (e.g., private semantic retrieval) [Arxiv'26, Spruce].
-
Cloud and AI infrastructure. Modern systems bring together specialized devices to support AI workloads, but their interconnects and shared resources often limit performance and reliability. We study micro-level contention and scheduling (e.g., PCIe, memory, and RNICs) [MCSched, GPUWeaver, SwiftRDMA, SkyRDMA]; coordinate hardware within servers and cloud gateways [Conspirator, CubeTrace]; and improve data movement across servers [ObjectCache, P-ECMP, MegaTE].
I am always looking for self-motivated students to work with. Read more here.
News
- 05/2026 CubeTrace, XFir, and Horizon are accepted by ACM SIGCOMM'26!
- 12/2025 I will serve on the TPC of SIGCOMM'26, EuroSys'27, and IEEE S&P'27, and as Publication Chair of APNet'26. Submissions are welcomed!
- 12/2025 Canopy, Obsidian, and Janus are accepted by USENIX NSDI'26!
- 08/2025 SwiftRDMA has won the Best Paper Award of APNet'25!
- 08/2025 I will serve on the TPC of CCS'26, IMC'26, and PETS/PoPETs'26, and ERC for MLSys'26. Submissions are welcomed!
- ......
Last Updated: September 2026