Photo

Zhizhen Zhong

Postdoctoral Associate
MIT Schwarzman College of Computing, Massachusetts Institute of Technology
Email:  zhizhenz [at] mit (dot) edu

Google Scholar | CSAIL Profile | ResearchGate | LinkedIn | GitHub | CV

Methodologies: Optics & Photonics | Computer Systems & Networking | Compiler & HW/SW Interface
Impact Areas: Reconfigurable Optical Networks | Photonic Computing Systems

Research

overview
I am building the next-generation high-performance distributed and cloud networked systems for data-intensive applications, such as large language models and big data analytics. Towards this vision, my research takes a cross-layer approach to co-design different stacks of the system: from novel photonic-electronic computing accelerators and networking devices, to reconfigurable hardware abstractions and kernel primitives, to system resource allocation, scheduler, and compiler software.

My work lies in the crossroads of "Networked Computer Systems" and "Optoelectronics & Silicon Photonics". My Research Statement describes my long-term vision towards Rearchitecting Data Movement in Next-Generation Networked Distributed Systems. There is also a Spotlight Article featuring my research and teaching work at MIT CSAIL.

Some Highlights are:
  • In-Network Computing: Accelerated Datapaths with Photonic-Electronic Co-Design
    • Lightning [SIGCOMM 2023]: the first smartNIC that serves DNN inference packets with a runtime-reconfigurable hardware datapath across photonic matrix multiplication cores and digital processing components.
    • IOI [OptSys 2021], [HotNets 2023]: the first smart transceiver for photonic computing inside the optical networks.
    • NetCast [Science 2022], [JLT 2024]: a server-client broadcasting network that splits photonic deep learning computation over deployed telecom fiber links.
  • Data Center Networks: Reconfigurable Topology with Optical Circuit Switching
    • TopoOpt [NSDI 2023]: a reconfigurable DNN training cluster in data centers that optimizes network topology and DNN parallelization strategy.
    • OTSS [OFC 2019], [OFC 2018]: joint allocation of computation resource and optical transmission time slices to realize ultralow-latency optical interconnect in HPC systems.
  • Wide-Area Networks: Cross-Layer (Application/IP/Optical) Resource Allocation and Traffic Engineering
    • MegaTE [SIGCOMM 2024]: the first TE system that manages fine-grained flows between millions of VMs or containers.
    • FlexWAN [SIGCOMM 2023]: a hardware-software optical system that supports reconfigurable fiber spectrum allocation for wide-area networks.
    • RSS [NSDI 2021]: risk-driven WAN controller at Facebook.
    • Arrow [SIGCOMM 2021]: the first cross-layer TE system that recovers from fiber cuts by reconfiguring the light to healthy fibers.
    • Bow [OFC 2021 Postdeadline]: an optical layer controller that configures optical amplifiers with Bayesian Optimization.

Teaching

MIT 6.S918: Optical Computing in the Era of AI (with Dr. Saumil Bandyopadhyay)
In this constantly evolving landscape of technology, light-based computing, commonly referred to as "optical" or "photonic" computing, is a revolutionary paradigm shift promising higher computing frequency and less energy consumption than traditional digital computing. This new course, offered at MIT EECS, aims to introduce students to this exciting and rapidly growing field, blending topics from photonics and optics to computer networks and systems. 1) How can light be used for computing, and why should we build optical computing hardware? 2) What are the fundamental devices used for photonic computing? 3) How do these photonic devices integrate into modern computer systems (both hardware and software) to serve real-world computing workloads? For more information, please refer to the course website: https://light-computing.github.io/

Videos

     

Publications ( ⚑ Preprints | ★ Selected | 📚 Positions | 📋 Full List )

Open Office Hour

I commit 1-2 hours of my spare time every week as my open office hour to meet, advise, mentor and help students/researchers who belong to underrepresented groups in the Computer Systems/Networking and Optics/Photonics community. Please fill out this form if you are interested.

Press Coverage

  • [ACM SIGCOMM 2023] First Photonic-Electronic Computing SmartNIC for Real-Time AI
    •       MIT News | Innovation Origins | Semiconductor Engineering | TechXplore
  • [Science 2022] Delocalized Photonic Deep Learning on the Internet's Edge
    •       MIT News | Optics & Photonics News | TechXplore
  • [ACM SIGCOMM 2021] ARROW: Restoration-Aware Traffic Engineering
    •       ACM TechNews | MIT News | TechXplore | Lifewire | 21st Century Tech | Future Tech
  • [NSDI 2021] Risk-Driven IP-Optical Network Management
    •       Facebook Engineering Blog
  • [CLEO-PR/OECC 2017] Sub-Wavelength Optical-Circuit-Switched Networks
    •       IEEE Photonics Society News

    Selected Talks

  • Lightning: A Reconfigurable Photonic-Electronic SmartNIC for Fast and Energy-Efficient Inference
    •      January 2024, Google Tech Talk, Google, USA
         October 2023, Stanford System Seminar, Stanford University, USA
         October 2023, Berkeley NetSys Seminar, UC Berkeley, USA
         September 2023, ACM SIGCOMM 2023, New York City, USA | Website | Video
  • ARROW: Restoration-Aware Traffic Engineering
    •      February 2022, Google Networking Research Summit, online | Website
         August 2021, ACM SIGCOMM 2021, online | Website | Video
  • ARROW and BOW: Bayesian Optimization for Reconfigurable Wide-Area Networks
    •      June 2021, OFC 2021, online | Website | Video
         June 2021, MIT and University of Washington Workshop on AI Implementations and Applications, online | Website | Video
  • Leveraging Optical Noise for Fast Fiber Cut Recovery
    •      August 2020, ACM SIGCOMM 2020 Workshop on Optical Systems Design (OptSys 2020), online | Website
  • The Impact of Social Distancing on Facebook's Optical Backbone
    •      August 2020, ACM SIGCOMM 2020 Workshop on Optical Systems Design (OptSys 2020), online | Website
  • Traffic-Driven Self-Ddaptive Optical Networking
    •      October 2019, Special Seminar, China Telecom Beijing Research Institute, Beijing, China
         July 2019, Special Seminar, University College London
         May 2019, Ph.D. Defense, Tsinghua University, Beijing, China
         March 2019, Special Seminar, Nokia Bell Labs (Crawford Hill), Holmdel, NJ, USA | Slides
  • Exploiting physical-layer flexibility for demand-responsive networks
    •      April 2018, Special Seminar, Georgia Institute of Technology, Atlanta, GA, USA | Slides
         March 2018, Post-OFC Workshop, University of California Davis, Davis, CA, USA | Website | Slides
  • Req2Route: learning to route for inter-domain lightpath provisioning
    •      March 2019, Post-OFC Workshop, University of California Davis, Davis, CA, USA | Website | Slides
         October 2018, AUA Postgraduate Forum, Tsinghua University, Beijing, China | Slides

    Professional Services

  • Organization
    •      Vice Chair, Optica Technical Group on Optical Communications
  • Conference
    •      Program Committee, USENIX NSDI 2025
         Chair, HotOptics 2024
         Program Committee, IEEE ICNP 2024
         External Reviewer, USENIX NSDI 2024
         Artifact Evaluation Committee Co-Chair, ACM SIGCOMM 2023
         Program Committee, ACM SIGCOMM 2023
         Program Committee, USENIX NSDI 2023
         Artifact Evaluation Committee, ACM SIGCOMM 2022
         External Reviewer, ACM IMC 2022
         Program Committee, ACM SIGCOMM 2021 Workshop on Optical Systems (OptSys 2021)
         Artifact Evaluation Committee, ACM SIGCOMM 2021
  • Journal
    •      Reviewer, IEEE/ACM Transactions on Networking
         Reviewer, IEEE Communications Magazine
         Reviewer, Optics Express
         Reviewer, Photonics Research
         Reviewer, IEEE/Optica Journal of Lightwave Technology
         Reviewer, IEEE/Optica Journal of Optical Communications and Networking
         Reviewer, IEEE Transactions on Network and Service Management
         Reviewer, IEEE Photoics Journal
         Reviewer, IEEE Access
         Reviewer, ACM SIGCOMM Computer Communication Review (SIGCOMM CCR)
         Reviewer, Photonic Network Communications
         Reviewer, Optical Switching and Networking

    Short Bio

    I am a postdoctoral researcher at the MIT Schwarzman College of Computing. Before joining MIT, I was a research scientist at Meta (Facebook) working on reconfigurable fiber-optics network infrastructures. I received my PhD in EE from Tsinghua University in July 2019. I was also a visiting PhD student in CS at the University of California Davis. My PhD thesis is on "Traffic-Driven Self-Adaptive Optical Networking" that designs reconfigurable optical networks for spatiotemporal traffic demand dynamics in both data centers and wide-area networks. I completed my Bachelor's Degrees in EE and Economics from Tsinghua University in 2014 and 2016, respectively. My work has been recognized by the Best Paper Award from OECC conference in 2022, the 1st Place Best Poster Award (1 out of 600+ posters) on IEEE CLEO-PR/OECC/PGC conference from IEEE Photonics Society in 2017.

    Miscellaneous

  • Photography
    •      In my spare time, I enjoy traveling and capturing the frozen time. See some of my photographical publications (peer reviewed) here: 1x

    All rights reserved, Zhizhen Zhong, MIT.

    Last updated: May, 2024.

    Accessibility