Zheng Yan
Member Of Technical Staff at OpenAI
California, United States
Summary
🤩
Rockstar🎓
Top SchoolZheng Yan is a seasoned software engineer with 13 years of experience designing and shipping high-performance ML inference and recommendation systems, now a Member of Technical Staff at OpenAI in California. At Meta he led LLM inference and RecSys runtime efforts, enabling serving for models with trillions of parameters and building seconds-level freshness guarantees on modern GPUs. His work spans low-level GPU communication and large-scale serving—he helped outperform NCCL for cluster training and scaled TorchRec inference components, contributing upstream to the popular pytorch/torchrec project. Comfortable across back-end, MLOps, and systems engineering, he pairs rigorous academic training with practical production impact and a recurring focus on quantization, batching, and device-aware inference optimizations. An understated strength is his track record of turning research ideas into deployable infra that measurably improves latency and freshness at massive scale.
13 years of coding experience
8 years of employment as a software developer
The Second High School Attached to Beijing Normal UniversityMaster of Science (M.S.) Computer Science, Master of Science (M.S.) Computer Science at University of MarylandBachelor of Engineering (B.Eng.) Computer Science, Bachelor of Engineering (B.Eng.) Computer Science at Zhejiang University
Chinese, English
Github Skills (13)
quantization10
pytorch10
c-language10
recommendation-system10
inference10
deep-learning10
cprogramming-language10
python10
optimisation10
optimization10
gpu9
mlops9
cuda9
Programming languages (5)
Github contributions (5)
pytorch/torchrec
Nov 2021 - Sep 2022
Pytorch domain library for recommendation systems
Role in this project:
Back-end Developer & MLOps Engineer
Back-end Developer & MLOps EngineerContributions:2 reviews, 35 commits, 43 PRs in 10 months
Contributions summary:Zheng made several commits related to the `torchrec` library, which focuses on recommendation systems. Their primary contributions involved addressing quantization-related issues, including fixes and improvements for quantized embedding modules and kernels. They also worked on enhancing the inference modules, such as improving the PredictModule and adding support for batching and device management in the inference pipeline. Furthermore, the user contributed to migrating Python modules and optimizing code for improved performance within the inference backend.
cudapytorchrecommendation-systemsdeep-learninggpu
zyan0/pytorch
Mar 2021 - Feb 2024
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:52 pushes, 12 branches in 2 years 11 months
pythongpu-accelerationdeep-learninggpuacceleration
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.