When Should an Agent Restart?
Two studies use large-scale trajectory replay and decision-value auditing to examine failure prediction and restarts for coding agents.
Read noteGuangba Yu.
Postdoctoral ResearcherThe Chinese University of Hong Kong
I’m Guangba Yu. My research focuses on the reliability and optimization of cloud and AI systems.
About me
Since July 2024, I have been a postdoctoral researcher at ARISE, The Chinese University of Hong Kong, mentored by Prof. Michael R. Lyu. My journey in tech was forged at Sun Yat-sen University, where I earned my Ph.D. and M.Eng under the guidance of Prof. Pengfei Chen. From 2022 to 2023, I was a research intern at Tencent’s WeChat Group through the CCF–Tencent Rhino-Bird Elite Program.
My path began not with code, but with traditional Chinese medicine. I started my academic career at Guangzhou University of Chinese Medicine, often called the "MIT of Traditional Chinese Medicine", with the dream of becoming a physician who diagnosed the "bugs" of human body. Today, I've channeled that diagnostic spirit from human systems to software systems, dedicating myself to identifying and healing the faults that affect software reliability and keeping our digital world healthy.
😃 Passionate about prescribing solutions for system reliability. If you're interested in building more robust software, let's connect!
My path began not with code, but with traditional Chinese medicine. I started my academic career at Guangzhou University of Chinese Medicine, often called the "MIT of Traditional Chinese Medicine", with the dream of becoming a physician who diagnosed the "bugs" of human body. Today, I've channeled that diagnostic spirit from human systems to software systems, dedicating myself to identifying and healing the faults that affect software reliability and keeping our digital world healthy.
😃 Passionate about prescribing solutions for system reliability. If you're interested in building more robust software, let's connect!
News
🏆 Our paper PreServe received the ACM SIGSOFT Distinguished Paper Award at ICSE 2026. CUHK announcement ↗
🏆 Our paper iKnow received the ACM SIGSOFT Distinguished Paper Award at ASE 2025. CUHK announcement ↗
🎉 Our Alert Management and Cloud Ops RAG Paper has been accepted by ASE'25.
Earlier news
🎉 Our Network Telemetry RCA Paper has been accepted by ToN. Congras to Hongyang!
🎉 Our Cloud Incident Management Paper has been accepted by TSE. Congras to Zilong!
🎉 Our Survey on Failure Analysis and Fault Injection in AI Systems has been accepted by TOSEM.
🎉 Log-based LLM training failures diagnosis framework L4 has been accepted by FSE'25. Congras to Zhihan!
🎉 Code knowledge enhanced root cause analysis framework CoCA has been accepted by ICSE'25. Congras to Yichen!
🎉 Microservice Trace Framework Mint has been accepted by ASPLOS'25. Congras to Haiyu!
🎉 FaaS configuration optimization Paper FaaSConf has been accepted by ASE'24 (118/587). Congras to Yilun!
🏆 Trace sampling paper TraStrainer has won Best Paper Award at FSE'24.
🎉 Request-level fault injection paper MicroFI has been accepted by TDSC. Congras to Hongyang!
🎉 Root cause change analysis paper ChangeRCA and trace sampling paper TraStrainer have been accepted by FSE'24 (56/483).
🎉 Multi-modal observability data RCA paper Nezha and database configuration optimization paper Diagconfig has been accepted by FSE'23.
Recent updates
Paper DigestAll digests
Tech NotesAll tech notes
How vLLM supports sparse attention and cross-layer KV sharing
vLLM's Mistral Large 4 support shows how model structure and serving infrastructure jointly affect long-context performance: sparse indexing reduces attention reads, cross-layer sharing avoids cache writes, and the runtime must handle parallelism, graph execution and disaggregated transfers.
Tensor Core acceleration for interatomic potentials with stability checks
NVIDIA approximates FP32 with lower-precision products and checks errors in trajectories and structural relaxation.
What I work on
Research areas
Education
Sun Yat-sen University
Ph.D. in Computer Science and Technology
Advisor: Prof. Pengfei Chen
Sun Yat-sen University
M.Eng. in Computer Technology
Advisor: Prof. Pengfei Chen
Guangzhou University of Chinese Medicine
B.Eng. in Medical Information Engineering
Advisor: Prof. Xiufeng Liu
Honors & awards
- ACM SIGSOFT Distinguished Paper Award, ICSE 2026 (PreServe)
- First Prize, Science & Technology Innovation Award, Guangdong Association for Promotion of S&T Achievement Transformation
- Third Prize, Science & Technology Progress Award, Chinese Institute of Electronics
- ACM SIGSOFT Distinguished Paper Award, ASE 2025 (iKnow)
- APSEC 2025 Distinguished Reviewer Award
- ACM China (Zhuhai) Doctoral Dissertation Award
- ACM SIGSOFT Distinguished Paper Award, FSE 2024 (TraStrainer)
- Best Cloud Computing Application Award, IEEE–CCF (LogReducer)
- Outstanding Graduate, Sun Yat-sen University
- National Scholarship, Ministry of Education of China
- CCF–Tencent Rhino-Bird Elite Talent Program & Special Scholarship
Academic service
- Program Committee: FSE 2027; FSE 2026 (IVR); ICSME 2026 (TDDS & VER); APSEC 2025; EuroSys 2025 (Shadow)
- Artifact Evaluation: OSDI 2022; USENIX ATC 2022
- Reviewer: TSE, TOSEM, TSC, TPDS, TC, TMC, KDD 2026, WWW 2024, PR, EAAI, TCC, JSEP, JSS, EMSE, Computer Networks, J Supercomput