• Programmable Network

  I have worked on newly emerged programmable network hardware since 2020, including programmable switching ASIC, SmartNIC, etc. I'm familiar with the design strategy and implementation of various programmable offloading systems. I also have some knowledge about software acceleration techniques, e.g. eBPF, DPDK.

• High Performance Network

  I have some experience in RDMA and high performance network. I joined lots of discussions about RDMA applications in industrial scenarios.

• Distributed ML Acceleration

  I've learned a lot about the state-of-the-art acceleration techniques of data-parallel distributed ML training, including on-switch aggregation, gradient compression, communication library optimization, etc.

• MulTA: Enabling Cost-efficient Multi-task Network Traffic Analysis. [GlobeCom 2023]

  Cheng Guo, Menghao Zhang, Guanyu Li, Mingwe Xu. CCF-C/TH-CPL-B, full paper acceptance ratio: Less than 40%

• IMap: Fast and Scalable In-Network Scanning with Programmable Switches. [NSDI 2022]

  Guanyu Li, Menghao Zhang, Cheng Guo, Han Bao, Mingwe Xu, Hongxin Hu, Fenghua Li. CCF-A/TH-CPL-A, full paper acceptance ratio: 50/298 = 16.8

• Switches are Scanners Too! A Fast and Scalable In-Network Scanner with Programmable Switches. [HotNets 2021]

  Guanyu Li, Menghao Zhang, Cheng Guo, Han Bao, Mingwe Xu, Hongxin Hu. CCF-C/TH-CPL-B, full paper acceptance ratio: 31/101 = 30.7

• 徐明伟,郭诚,李冠宇,张梦豪,王士诚,李琦. 流量特征提取方法、系统、存储介质及电子设备[P]. CN114024758B, 2022.12.23

• 徐明伟,张梦豪,李冠宇郭诚,包涵. 网络扫描装置. CN115314425A, 2022.11.08

Programming

• Proficient

  C/C++
  Python
  P4
  
• Competent

  Bash
  Matlab
  LaTeX
  HTML
  
• Beginner

  Java
  HDL
  

Software & Tools

• Proficient

  Matplotlib
  Office
  
• Competent

  Numpy
  TensorFlow
  Keras
  Sklearn
  Git
  VM
  
• Beginner

  Docker
  

Language

• Proficient

  Chinese
  English
  
• Competent

  Japanese
  
• Beginner

  German