Longjie Zheng
Software Engineer II at Prime Video & Amazon MGM Studios
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolLongjie Zheng is a Software Engineer II with six years of experience building and optimizing ML deployment and acceleration systems, currently at Prime Video & Amazon MGM Studios in New York. He holds a B.E. from SJTU and an MCS from UIUC, and has worked on compiler-style optimizations, operator fusion, and quantization for production AI workloads. At Hugging Face he contributed to PyTorch 2.0 compilation support for LLMs and automatic model parallelism, and at SenseTime he led an open-source AI acceleration toolkit that attracted 1k+ GitHub stars. His open-source contributions to PPQ (PPL Quantization Tool) include ONNX Runtime/MetaX fixes and adding ops like Mod and Softplus, reflecting deep expertise in model conversion and quant-aware training. He blends research-driven experimentation with production engineering, often turning recent academic techniques into deployable tooling. Colleagues rely on him to bridge large-model research and robust deployment pipelines.
6 years of coding experience
1 year of employment as a software developer
Master's degree, Computer Science, Master's degree, Computer Science at University of Illinois Urbana-ChampaignChongqing No.1 Middle SchoolBachelor of Engineering - IEEE Honor class, Computer Science, Bachelor of Engineering - IEEE Honor class, Computer Science at Shanghai Jiao Tong University
Github Skills (7)
neural-network10
quantization10
pytorch10
deep-learning10
onnx10
caffe8
cuda7
Programming languages (3)
Github contributions (5)
OpenPPL/ppq
Jan 2022 - Jun 2022
PPL Quantization Tool (PPQ) is a powerful offline neural network quantization tool.
Role in this project:
ML Engineer
ML EngineerContributions:20 commits, 24 PRs, 3 comments in 5 months
Contributions summary:Longjie primarily contributed to the PPQ (PPL Quantization Tool) repository, focusing on enhancing its capabilities for neural network quantization. Their work included fixing issues related to ONNX Runtime and MetaX export, indicating expertise in model conversion and deployment. The user added support for new features such as 'Mod' and 'Softplus' operations, demonstrating expansion of the tool's functionality, and they also refined training algorithms, streamlining the model quantization process.
cudapytorchdeep-learningonnxnetwork-quantization
zhenglongjiepheonix/transformers
May 2024 - Sep 2024
Contributions:69 pushes, 7 branches in 4 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.