Jie Sun
SWE at Google
Pleasanton, California, United States
Summary
🤩
Rockstar🎓
Top SchoolJie Sun is a seasoned software engineer with 16 years of experience building high-performance systems for Google and NVIDIA, currently based in Pleasanton, California. He specializes in ML compilers, accelerator co-design, and profiling—owning TPU/GPU profiling, tracing support, and large-scale ML tooling that reportedly saved billions in infrastructure costs. Jie blends deep backend and systems expertise (XLA, PjRT, JAX, TPU/GPU codegen and memory/alias analysis) with full-stack visualization work on projects like TensorBoard, and has contributed to flagship open-source projects such as TensorFlow and TensorBoard. His work spans from low-level device simulation and compiler internals to frontend profiling UI, plus production services like fleetwide accelerator profiling and ML symbol servers. Notably, he has driven practical innovations for LLM parallelism modeling and PCIe offloading for KV cache, demonstrating both hardware-aware optimization and software ergonomics. A Peking University computational science graduate, Jie pairs rigorous academic training with decades of production-grade engineering across hardware and ML infrastructure.
16 years of coding experience
7 years of employment as a software developer
Master's Degree Computational Science, Master's Degree Computational Science at Peking University
Github Skills (20)
javascript10
c-language10
python10
memory-management10
tensorflow10
profiling10
html10
xla10
compiler10
cprogramming-language10
protocol-buffers9
data-visualizations9
data-visualisation9
polymer9
linear-algebra9
Programming languages (3)
Github contributions (5)
tensorflow/tensorboard
Nov 2017 - May 2018
TensorFlow's Visualization Toolkit
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:6 commits, 14 PRs, 52 comments in 6 months
Contributions summary:Jie primarily contributed to the TensorBoard plugin for profiling TensorFlow models, focusing on the input pipeline analysis. Their work included implementing a new analyzer for TPU input pipelines, fixing bugs in the existing profiling tools, and supporting streaming trace viewer functionality. They also improved the UI of the analyzer by adding captions and formatting the x-axis. The changes involved both front-end (HTML/JavaScript) and back-end (Python) work related to data visualization and performance analysis.
tensorflowvisualization
tensorflow/tensorflow
Aug 2022 - Jan 2023
An Open Source Machine Learning Framework for Everyone
Role in this project:Back-end Developer
Back-end DeveloperContributions:11 reviews, 40 commits, 1 PR in 5 months
Contributions summary:Jie's commits primarily focused on refactoring and enhancing the memory visualization utilities within the TensorFlow project. Their contributions included modifying the code for better logical buffer handling, separating related information, and introducing improvements to the display and management of memory allocation timelines. These changes involved updating data structures, optimizing memory usage tracking, and streamlining the code related to the heap simulator trace.
machine-learningtensorflowpythondeep-learningdeep-neural-networks
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.