Jiong Gong
Senior Tech Expert at Huawei
Shanghai, Shanghai, China
Summary
🤩
Rockstar🎓
Top SchoolJiong Gong is a Senior Tech Expert based in Shanghai with nine years focused on deep learning framework optimization and a two-decade engineering pedigree at Intel and now Huawei. He has driven architecture and implementation across frontend APIs, operator optimization, graph compilers and accelerator-aware libraries, and is a noted contributor and maintainer within the PyTorch community—working on inductor/autograd improvements, memory-efficient in-place buffers, profiling of C++ kernels, and fixes that improved model accuracy and performance. At Intel he led efforts in low-precision and DL Boost technologies and helped productionize CPU-optimized PyTorch (IPEX); earlier roles ranged from UEFI firmware engineering to managing AI-driven mobile test automation. Known for translating compiler- and kernel-level innovations into tangible speed and memory gains, he blends deep systems-level know-how with practical production delivery.
9 years of coding experience
19 years of employment as a software developer
Shanghai High SchoolShanghai Jiao Tong University
Github Skills (16)
performance-tuning10
pytorch10
performance-monitor10
c-language10
performance-analysis10
performance-analytics10
performance-monitoring10
cprogramming-language10
vectorization10
performance-measurement10
machine-learning9
deep-learning9
cuda9
tensor8
autograd8
Programming languages (8)
Github contributions (5)
pytorch/pytorch
Aug 2018 - Jan 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
ML Engineer
ML EngineerContributions:2971 reviews, 76 commits, 124 PRs in 4 years 6 months
Contributions summary:Jiong's commits focused on enhancing the PyTorch framework, particularly around its inductor and autograd functionalities. The contributions involved enabling and optimizing in-place buffer operations, which improved memory efficiency. They added support for recording individual C++ kernel execution times in the PyTorch profiler and introduced options to mark the duration of wrapper calls. Further work included fixing buffer overflows in bfloat16 interpolation, enhancing the efficiency of vectorization, and improving accuracy across multiple models.
pythongpu-accelerationdeep-learninggpunumpy
jgong5/pytorch
Aug 2018 - May 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:92 pushes, 30 branches in 4 years 10 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.