Hailiang Zhao
Scholar

Hailiang Zhao

Google Scholar ID: wMj2rFsAAAAJ
ZJU 100 Young Professor, Zhejiang University
Service ComputingEdge ComputingLearning-Augmented Algorithms
Citations & Impact
All-time
Citations
1,907
 
H-index
13
 
i10-index
13
 
Publications
20
 
Co-authors
4
list available
Resume (English only)
Academic Achievements
  • Published several papers including 'OmniFuser: Adaptive Multimodal Fusion for Service-Oriented Predictive Maintenance', 'Agentic Services Computing', 'Toward Robust and Efficient ML-Based GPU Caching for Modern Inference', etc. Some of these papers have been included in arXiv preprints or will be presented at NeurIPS '25.
Research Experience
  • ZJU 100 Young Professor at School of Software Technology, Zhejiang University. Research areas include Learning-Augmented Algorithms & Systems and Industrial Intelligence. Representative contributions include the PFSUM algorithm for the classic Bahncard problem (NeurIPS '24) and a robustification framework Guard for the online caching problem (NeurIPS '25). These algorithms are designed not only for theoretical guarantees but also for real-world computer systems such as MLSys, microservices platforms, and heterogeneous job schedulers. Additionally, exploring embedding these techniques into system control loops to support predictive autoscaling, dynamic load dispatching, and low-latency request handling under highly non-stationary workloads.
Education
  • PhD in College of Computer Science and Technology at Zhejiang University (ZJU) in June 2024, supervised by Prof. Shuiguang Deng. Visiting PhD student at PDCL Lab, Nanyang Technological University (NTU) from September 2022 to September 2023, under the supervision of Prof. Xueyan Tang.
Background
  • Research Interests: Learning-Augmented Algorithms & Systems, Industrial Intelligence. Focuses on integrating machine learning predictions into classic algorithmic frameworks to improve decision-making in uncertain and dynamic environments; explores how to unify temporal, structured, and perceptual signals from discrete manufacturing processes to enable adaptive, learning-driven decision making.
Miscellany
  • Interested students who would like to pursue a Master's or Ph.D. under my supervision are welcome to contact me.