Man-Kit Sit

me.jpeg

I am a PhD student at the University of Edinburgh, under the supervision of Luo Mai. Prior to that, I received a MSc in Computer Science from Keio University in 2018 and a BEng in Computer Engineering from the Chinese University of Hong Kong. My research interests revolve around the intersection of Large Scale Machine Learning, Distributed Systems and Specialized Accelerators. The current research endeavors are centered on the design and development of automatic and scalable distributed systems to train gigantic machine learning models efficiently.

Research Interests

  • Machine Learning Systems
  • Distributed Systems
  • Computer Architecture
  • Reconfigurable Computing (or FPGA)
  • Large-Scale Reinforcement Learning
  • Blockchain

Publications

2023

  1. ICML
    GEAR: A GPU-Centric Experience Replay System for Large Reinforcement Learning Models
    Hanjing Wang*, Man-Kit Sit*, Congjie He, Ying Wen, Weinan Zhang, Jun Wang, Yaodong Yang, and Luo Mai
    In 2023 International Conference on Machine Learning, Jul 2023
  2. arXiv
    Quiver: Supporting GPUs for Low-Latency, High-Throughput GNN Serving with Workload Awareness
    Zeyuan Tan, Xiulong Yuan, Congjie He, Man-Kit Sit, Guo Li, Xiaoze Liu, Baole Ai, Kai Zeng, Peter Pietzuch, and Luo Mai
    Jul 2023

2022

  1. OSDI
    Ekko: A Large-Scale Deep Learning Recommender System with Low-Latency Model Update
    Chijun Sima, Yao Fu, Man-Kit Sit, Liyi Guo, Xuri Gong, Feng Lin, Junyu Wu, Yongsheng Li, Haidong Rong, Pierre-Louis Aublin, and Luo Mai
    In 16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22), Jul 2022

2020

  1. arXiv
    Towards Improving the Performance of BFT Consensus For Future Permissioned Blockchains
    Manuel Bravo, Zsolt István, and Man-Kit Sit
    CoRR, Jul 2020

2019

  1. FCCM
    Towards Efficient Deep Neural Network Training by FPGA-Based Batch-Level Parallelism
    Cheng Luo, Man-Kit Sit, Hongxiang Fan, Shuanglong Liu, Wayne Luk, and Ce Guo
    In 2019 IEEE 27th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM), Jul 2019

2017

  1. FPT
    FPGA-based Accelerator for Losslessly Quantized Convolutional Neural Networks
    Man-Kit Sit, Ryosuke Kazami, and Hideharu Amano
    In 2017 International Conference on Field Programmable Technology (ICFPT), Jul 2017