Shao-han Tang
Software Engineering Manager at Zoox
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolShao-han Tang is a Software Engineering Manager in the San Francisco Bay Area with 11 years of experience building autonomous vehicle frameworks, robotic middleware, and camera pipelines. At Zoox he leads software-defined vehicle and autonomous software architecture, bringing deep systems-level C++ expertise from prior roles at Qualcomm where he drove camera HAL3 and pipeline re-architecture for flagship devices. He pairs hands-on performance engineering—contributions to high-profile open-source projects like Triton, PyTorch, and karpathy’s llm.c—with managerial responsibilities, improving CI, CUDA kernels, and documentation to make complex ML and robotics codebases more reliable and readable. Known for “plumbing” large software pipelines end-to-end, he combines low-level debugging chops with a knack for clarifying developer experience, evidenced by tutorial and CI improvements across repositories. He holds an MS in Computer Science from USC and a BS in Electrical and Electronics Engineering from National Taiwan University.
11 years of coding experience
9 years of employment as a software developer
Master of Science (M.S.), Computer Science, Master of Science (M.S.), Computer Science at University of Southern CaliforniaBachelor of Science - BS, Electrical and Electronics Engineering, Bachelor of Science - BS, Electrical and Electronics Engineering at National Taiwan University
English, Chinese
Stackoverflow
3reputation
47reached
0answers
1question
Github Skills (50)
pytorch10
algorithms10
github-ci10
develop10
c-language10
python10
gpu-programming10
solve10
machine-learning10
c1110
data-structure10
cicd10
c1710
solution10
kernel10
Programming languages (7)
Github contributions (5)
pytorch/examples
Dec 2023 - Feb 2025
A set of examples around pytorch in Vision, Text, Reinforcement Learning, etc.
Role in this project:
MLOps Engineer
MLOps EngineerContributions:1 review, 21 PRs, 2 pushes in 1 year 1 month
Contributions summary:Shao-han focused on improving the continuous integration (CI) and build process for the example projects. Their work included adding CI jobs for various C++ examples, including autograd, custom-dataset, MNIST, regression, and dcgan. They also addressed minor issues in the examples, such as fixing dataset URLs and ensuring code consistency. Finally, they enhanced the dcgan example with command-line argument parsing and code readability improvements.
pytorchreinforcement-learning
karpathy/llm.c
Apr 2024 - Sep 2024
LLM training in simple, raw C/CUDA
Role in this project:ML Engineer
ML EngineerContributions:5 reviews, 61 PRs, 28 comments in 4 months
Contributions summary:Shao-han primarily contributed to optimizing and expanding the functionality of CUDA kernels for language model training. They addressed errors related to mathematical constants, improved the efficiency of existing softmax kernels, and introduced a new parallel softmax implementation. Furthermore, they fixed bugs in other CUDA kernels like matmul and layernorm, and addressed compiler warnings, demonstrating a focus on performance and accuracy within the LLM framework.
cudallm
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.