Hong Zhang
Assistant Professor of Cheriton School of Computer Science, University of Waterloo
Office: DC3530
I develop high-performance, scalable systems for big data and ML applications. My research advocates an application-oriented design principle for big data and ML systems: fully exploiting application-specific structures --- communication patterns, execution dependencies, ML model structures, etc. --- to suit application-specific performance demands. This principle has led to several scalable systems with theoretically sound scheduling algorithms tailored for different big data and ML applications. I received a Google Ph.D. Fellowship in systems and networking.
Biography: I was a PostDoc in the RISELab at UC Berkeley working with Prof. Ion Stoica. I received my Ph.D. from Department of Computer Science and Engineering, Hong Kong University of Science and Technology, where I worked with Prof. Kai Chen in System networkING (SING) research group.
Interests:
Large-scale Data Analytics
Distributed ML Training & Serving Systems
Application and Network Scheduling
Data Center Networking
Serverless Computing and Cloud Computing
news
Sep 17, 2024 |
I am looking for self-motivated PhD students to work with me on distributed systems (ML systems in particular) starting Spring/Fall 2025. Drop me an email if you are interested! |
---|---|
Sep 17, 2024 |
Serve on the program committee for OSDI'25, NSDI'25, and EuroSys'25. |
selected publications
-
___ASPLOS___RainbowCake: Mitigating Cold-starts in Serverless with Layer-wise Container Cachine and SharingIn ASPLOS 2024
-
___EuroSys___Accelerating Privacy-Preserving Machine Learning with GeniBatchIn EuroSys 2024
-
_____NSDI_____SHEPHERD: Serving DNNs in the WildIn Proceedings of the 20th USENIX Symposium on Networked Systems Design and Implementation, 2023
-
_____NSDI_____NetHint: White-Box Networking for Multi-Tenant Data CentersIn Proceedings of the 19th USENIX Symposium on Networked Systems Design and Implementation, 2022
-
__SIGCOMM__LiteFlow: Towards High-performance Adaptive Neural Networks for Kernel DatapathIn Proceedings of the ACM SIGCOMM 2022 Conference, 2022
-
_____NSDI_____Caerus: NIMBLE Task Scheduling for Serverless AnalyticsIn Proceedings of the 18th USENIX Symposium on Networked Systems Design and Implementation, 2021
-
__SIGCOMM__Resilient Datacenter Load Balancing in the WildIn Proceedings of the ACM SIGCOMM 2017 Conference, 2017
-
__SIGCOMM__CODA: Toward Automatically Identifying and Scheduling Coflows in the DarkIn Proceedings of the ACM SIGCOMM 2016 Conference, 2016
-
___EuroSys___Guaranteeing Deadlines for Inter-Datacenter TransfersIn Proceedings of the 10th European Conference on Computer Systems, 2015