Zhijian Liu
Mountain View, California, United States
Summary
🤩
RockstarZhijian Liu is a research scientist at NVIDIA and an incoming assistant professor at UCSD, bringing ten years of experience at the intersection of efficient machine learning and systems. He earned his PhD and S.M. in EECS from MIT after a B.Eng. in Computer Science from Shanghai Jiao Tong University, and spent formative years as a research assistant and intern at MIT and Toyota Research Institute. His work spans both theory and production: he has contributed practical, compatibility-focused improvements to widely used sparse-ML toolkits such as TorchSparse and SPVNAS, fixing edge-case sparse-crop behavior and modernizing code for newer frameworks. Based in Mountain View, he blends rigorous academic research with hands-on engineering to make sparse-convolution pipelines more robust and deployable.
11 years of coding experience
Github Skills (9)
acceleration10
pytorch10
deeplearning10
deep-learning10
python10
3d10
gpu-programming9
type-annotations8
computer-vision8
Programming languages (5)
Github contributions (5)
mit-han-lab/torchsparse
Sep 2020 - May 2022
[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.
Role in this project:
ML Engineer
ML EngineerContributions:5 releases, 92 reviews, 42 commits in 1 year 7 months
Contributions summary:Zhijian contributed to the `torchsparse` repository, a framework for sparse convolution on GPUs. Their work primarily involved minor code improvements and refactoring. Specifically, they removed unnecessary imports, added type annotations for the `KernelRegion` class, and reformatted the `setup.py` file. The user also fixed a sparse crop functionality.
pytorchinference-enginedeep-learningaccelerationinference
mit-han-lab/spvnas
Oct 2020 - Aug 2021
[ECCV 2020] Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
Role in this project:ML Engineer
ML EngineerContributions:11 reviews, 21 commits, 16 PRs in 9 months
Contributions summary:Zhijian primarily focused on enhancing the existing 3D deep learning framework. They updated the code to support newer versions of TorchSparse, a key library for sparse tensor operations, which is central to the project's functionality. They also addressed size mismatches and fixed typos within tutorial documentation, thus improving code compatibility and user experience.
kittieccvsparseconvolutionpoint-cloud
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.