Shigang Li
Research Professor at Beijing University of Posts and Telecommunications
Urbana, Illinois, United States
Summary
👤
Senior🎓
Top SchoolShigang Li is a research-focused professor and parallel computing expert with 11+ years of experience building high-performance and heterogeneous deep learning systems across academia and industry. Currently a Research Professor at Beijing University of Posts and Telecommunications after a postdoc at ETH Zurich, he specializes in parallel and distributed deep learning, GPU-accelerated computing, MPI, and high-performance computing. His work has earned multiple best/Outstanding Paper nominations and awards (SC, PPoPP, HPDC, MLSys) and he contributes to the community as an associate editor, conference publications chair, and program committee member. Comfortable bridging theory and practice, he has industrial experience from Baidu’s Institute of Deep Learning and a joint PhD background from UIUC, lending him both systems-building chops and rigorous research depth.
11 years of coding experience
9 years of employment as a software developer
University of Illinois Urbana-ChampaignDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of Science and Technology Beijing
Github Skills (42)
caffe10
chimera10
sparse-matrix10
machine-learning10
bidirectional10
deep-learning10
fpga10
programming-language9
vivado-hls9
gradient-descent9
transformers9
large-scale9
scale9
cuda9
parallel-computing8
Programming languages (6)
Github contributions (5)
ParCIS/Magicube
Oct 2021 - Nov 2022
Magicube is a high-performance library for quantized sparse matrix operations (SpMM and SDDMM) of deep learning on Tensor Cores.
Contributions:291 commits, 8 pushes in 1 year 1 month
deep-learningsparse-matrixtensor
ParCIS/Ok-Topk
Nov 2021 - Jun 2022
Ok-Topk is a scheme for distributed training with sparse gradients. Ok-Topk integrates a novel sparse allreduce algorithm (less than 6k communication volume which is asymptotically optimal) with the decentralized parallel Stochastic Gradient Descent (SGD) optimizer, and its convergence is proved theoretically and empirically.
Contributions:26 commits, 8 pushes, 4 comments in 6 months
distributed-traininggradient-descentdistributed-deep-learningsparse-allreducetopk-sgd
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.