Kaiqiang Xu
photo

Kaiqiang Xu

PhD in Computer Science (2025), HKUST

Research

My research focuses on improving the efficiency and accessibility of machine learning systems and AI computing infrastructure. A core theme of my work is designing new abstractions, parallelization strategies, and scheduling algorithms that exploit the unique characteristics of machine learning workloads and large-scale clusters.

I am currently in the final year of my PhD in Computer Science at HKUST, advised by Prof. Kai Chen. During my time at HKUST, I led the design and development of TACC, an AI cluster that supports over 600 university members. I earned my BSc in Computer Science from Renmin University of China.

My work has been published in top-tier conferences across computer networks, operating systems, architecture, and databases, including NSDI, OSDI, ASPLOS, and SIGMOD. I am also a visiting researcher at Google, where I work on efficient infrastructure and frameworks for large-scale models.

Selected Papers

USENIX NSDI, 2025 | USENIX Symposium on Networked Systems Design and Implementation
GREEN: Carbon-efficient Resource Scheduling for Machine Learning Clusters link download paper pdf file
Kaiqiang Xu, Decang Sun, Han Tian, Junxue Zhang, Kai Chen
ACM ASPLOS, 2025 | ACM International Conference on Architectural Support for Programming Languages and Operating Systems
Design and Operation of Shared Machine Learning Clusters on Campus link download paper pdf file
Kaiqiang Xu, Decang Sun, Hao Wang, Zhenghang Ren, Xinchen Wan, Xudong Liao, Zilong Wang, Junxue Zhang, Kai Chen
based on
TACC
ACM SIGMOD, 2025 | ACM International Conference on Management of Data
Sequoia: An Accessible and Extensible Framework for Privacy-Preserving Machine Learning over Distributed Data link download paper pdf file
Kaiqiang Xu, Di Chai, Junxue Zhang, Fan Lai, Kai Chen
USENIX OSDI, 2025 | USENIX Symposium on Operating Systems Design and Implementation
Enabling Efficient GPU Communication over Multiple NICs with FuseLink
Zhenghang Ren, Yuxuan Li, Zilong Wang, Xinyang Huang, Wenxue Li, Kaiqiang Xu, Xudong Liao, Yijun Sun, Bowen Liu, Han Tian, Junxue Zhang, Mingfei Wang, Zhizhen Zhong, Guyue Liu, Ying Zhang, Kai Chen

Full publication list available on Google Scholar.

Startups

My work seeks the synergy between technology and creativity. I aim to design exceptional user experiences in products that embed advanced technologies, transforming the way people interact with complex systems. The Wall Street Journal featured an exclusive interview with me for creating the world-recognized all-terrain robot HEXA.

I was named to the Forbes 30 Under 30 list twice, in 2020 (China) and 2022 (Asia).

COO, Vincross Robotics, 2016-2020
Investors include GGV Capital, Lenovo Capital and ZhenFund.
Vincross empowers users to access and build the future of robotics. I led a global team and made the first programmable all-terrain robot HEXA for individuals to create robotic applications. Reported by the Wall Street Journal, The Verge, WIRED, IEEE Specturm and more.

Earlier in my career, I worked in serveral startups including as an algorithm engineer at Zhihu.com (NYSE: ZH) and a backend engineer at Tantan (now part of MOMO, NASDAQ: MOMO). See my LinkedIn profile for more details.

Outside of work, I enjoy art and music whenever I can. My favorite artist is Edward Hopper for his Nighthawks (1942), among others. I recommend the timeless musical Les Misérables to everyone. I dance Argentine Tango and have a love for Golden Age Tango music. Besides my hometown of Jinan, I have also called Ann Arbor, Beijing, Berkeley, Shenzhen and Hong Kong home.