Sijia Chen
Member Of Technical Staff at OpenAI
Sunnyvale, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSijia Chen is a Member of Technical Staff in Sunnyvale with 10 years of experience building high-performance ML systems and GPU-optimized kernels. Previously a software engineer at Meta, Sijia drove attention and inference optimizations—FP8, MoE, and novel attention variants—bringing deep systems and kernel-level expertise to production-scale models. Their open-source contributions to PyTorch include nuanced fixes to embedding quantization, Triton kernel control-flow handling, and FBGEMM integration, reflecting an engineer who reads and improves core ML frameworks. Trained with an MS in Computer Engineering from Carnegie Mellon and a BS from Beijing University of Posts and Telecommunications, they combine rigorous academic foundations with practical production work. Colleagues describe them as a pragmatic optimizer who surfaces subtle correctness issues (like non-contiguous memory format bugs) that unlock faster, more reliable model inference. Now at OpenAI, Sijia continues to focus on squeezing performance and robustness from the stack where it matters most.
9 years of coding experience
7 years of employment as a software developer
Bachelor of Science - BS Computer Science and Technology, Bachelor of Science - BS Computer Science and Technology at Beijing University of Posts and TelecommunicationsMaster of Science - MS Computer Engineering, Master of Science - MS Computer Engineering at Carnegie Mellon University
Chinese, English
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (12)
quantization10
pytorch10
machine-learning10
triton10
python9
gpu9
deep-learning9
c-language8
cprogramming-language8
autograd7
tensor7
mlops6
Programming languages (3)
Github contributions (5)
pytorch/pytorch
May 2022 - Nov 2022
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
ML Engineer
ML EngineerContributions:17 reviews, 7 commits, 17 PRs in 6 months
Contributions summary:Sijia contributed significantly to the PyTorch repository, focusing on improving and fixing issues related to quantization, particularly embedding quantization and the use of FBGEMM. They addressed a bug in embedding quantization when memory formats were not contiguous. Furthermore, they worked on supporting float and handling `scf.for` and `scf.while` cases within the user-written Triton kernels, showing a deep understanding of the framework's inner workings, and also fixed an SDPA AOT export issue.
pythongpu-accelerationdeep-learninggpunumpy
sijiac/pytorch
May 2022 - Feb 2025
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:55 pushes, 12 branches in 2 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.