TJ Xu
Senior Software Engineer at Microsoft
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolJie Xu is a Senior Research Scientist at NVIDIA with nine years of experience bridging robotics, physics-based simulation, and machine learning. He earned a PhD in Computer Science from MIT and has a track record of research roles at MIT CSAIL, Tsinghua, CMU, and internships at NVIDIA and Snap, applying reinforcement learning and computational fabrication to real-world robotic systems. At NVIDIA he focuses on accelerating deep learning and low-level performance engineering—evidenced by contributions to TensorFlow that optimize GPU memory allocation and allreduce/collective operations for faster distributed training. Jie blends rigorous academic research with production-minded engineering, often reworking internal memory and communication primitives to shave latency at scale. He is based in Redmond, Washington and known for turning physics-aware simulation insights into practical ML and robotics performance wins. Colleagues describe him as a pragmatic problem-solver who surfaces non-obvious system-level improvements that materially improve throughput.
9 years of coding experience
8 years of employment as a software developer
Bachelor’s Degree, Computer and Electrical Engineer, Bachelor’s Degree, Computer and Electrical Engineer at University of Toronto
Github Skills (5)
ccl10
cuda10
memory-management10
c-language10
cprogramming-language10
Programming languages (6)
Github contributions (5)
tensorflow/tensorflow
Nov 2022 - Jan 2023
An Open Source Machine Learning Framework for Everyone
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:31 reviews, 4 commits, 3 PRs in 1 month
Contributions summary:TJ focused on optimizing memory allocation and improving performance within the TensorFlow framework, particularly in the context of GPU-accelerated machine learning. Their work involved modifying the per-process memory allocation to use total memory as a base, as well as addressing and fixing compile errors related to changes in code. The user also modified some internal functions for performance improvements. Additionally, the user contributed code to enhance the allreduce computation and the collective permutation to handle mem p2p more efficiently, showing a deep understanding of the underlying architecture.
machine-learningtensorflowpythondeep-learningdeep-neural-networks
Tixxx/xla
May 2023 - Jul 2026
A machine learning compiler for GPUs, CPUs, and ML accelerators
Contributions:5 PRs, 451 pushes, 99 branches in 3 years 2 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.