Vincent Zhang
GPU Algorithm Engineer at NVIDIA
Shanghai, China
Summary
🤩
Rockstar🎓
Top SchoolVincent Zhang is a GPU algorithm engineer with 10 years of experience building high-performance deep learning tooling and CV systems, currently working on Torch-TensorRT, CV-CUDA, TensorRT-LLM and Cutlass at NVIDIA. He previously led multiple object tracking research and model tooling efforts at Tencent Youtu Lab and contributed to TNN and related deployment tools. Vincent is an active open-source contributor to the PyTorch/TensorRT compiler, adding many operator converters and tests to broaden TensorRT’s ability to run PyTorch models efficiently on NVIDIA GPUs. With a strong academic foundation from Shanghai Jiao Tong University and Zhejiang University in electrical and electronic engineering, he blends research rigor with production-first engineering. Colleagues know him for turning tricky tensor/operator edge cases into robust converter implementations that unlock real-world model deployment.
10 years of coding experience
3 years of employment as a software developer
Bachelor's degree, Electronic and Information Engineering, Bachelor's degree, Electronic and Information Engineering at Zhejiang UniversityMaster's degree, Electrical and Electronics Engineering, Master's degree, Electrical and Electronics Engineering at Shanghai Jiao Tong University
English, Chinese
Github Skills (7)
tensorrt10
pytorch10
deep-learning10
python9
test-automation9
machine-learning9
cuda8
Programming languages (3)
Github contributions (5)
pytorch/TensorRT
Nov 2020 - Aug 2022
PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT
Role in this project:
ML Engineer
ML EngineerContributions:36 reviews, 78 commits, 23 PRs in 1 year 8 months
Contributions summary:Vincent contributed significantly to the PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT. Their work focused on adding support for various PyTorch operations (e.g., leaky_relu, squeeze/unsqueeze, gt/lt/eq/ge/le, true_divide, floor_divide, max, min, rsub, topk, erf/asinh/acosh/atanh, div.Scalar, mean with negative dim, transpose with negative dim, and arange) to the TensorRT converter. This involved writing converter implementations and test cases for each supported operator, enhancing the compiler's ability to translate more PyTorch models for optimized execution on NVIDIA GPUs.
cudapytorchjetsonnvidiadeep-learning
inocsin/hpcsimulator
Apr 2017 - Dec 2017
Contributions:42 commits, 35 pushes, 2 branches in 7 months
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.