Zhicheng Xiong
Deep Learning Performance Architect at NVIDIA
Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolZhicheng Xiong is a Deep Learning Performance Architect at NVIDIA with five years of hands-on experience accelerating machine learning systems and compiler-level automatic differentiation. Trained at Tsinghua with an exchange at ETH Zürich, he blends strong academic grounding with practical contributions to high-performance projects like Taichi and LLVM’s Clad AD plugin. His open-source work includes performance-focused enhancements to Taskflow—where his data abstraction and documentation efforts helped users leverage asynchronous tasking and CUDA algorithms—and a benchmarked scheduler that outperformed Intel oneTBB. Zhicheng’s background spans compiler plugins, NDArray-integrated autodiff, and GPU-aware optimization, enabling him to bridge algorithmic research and production-grade performance engineering. Known for shipping clear documentation alongside code, he brings a developer-centric mindset to complex systems design. Based in Beijing, he pairs low-level systems expertise with a taste for elegant, user-friendly tooling (motto: Cor Cordium).
5 years of coding experience
1 year of employment as a software developer
High School Diploma, High School Diploma at High School Affiliated to Anhui Normal UniversityBachelor of Engineering - BE, Bachelor of Engineering - BE at Tsinghua UniversityExchange Program, Exchange Program at ETH Zürich
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (4)
html10
documentation10
concurrent-programming8
cuda8
Programming languages (7)
Github contributions (5)
taskflow/taskflow
Aug 2022 - Aug 2022
A General-purpose Task-parallel Programming System using Modern C++
Role in this project:
Technical Writer
Technical WriterContributions:7 commits, 3 pushes in 6 days
Contributions summary:Zhicheng primarily focused on modifying documentation files related to asynchronous tasking and CUDA standard algorithms within the Taskflow repository. The commits involve changes to HTML files, indicating updates and improvements to the project's documentation. These changes suggest an effort to enhance clarity and comprehensiveness for users of the Taskflow library.
multithreadingcuda-programmingtask-programmingmulti-threadingconcurrent-programming
PhrygianGates/clad
Mar 2023 - Sep 2023
clad -- automatic differentiation for C/C++
Contributions:116 pushes, 8 branches in 5 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.