John Fu
Software Design Engineer at Microsoft
Redmond, Washington, United States
Summary
🤩
RockstarJohn Fu is a Software Design Engineer at Microsoft with nine years of experience building production-grade tooling for ML model deployment and profiling. Based in Redmond, he contributes to high-impact open-source projects like ONNX Runtime—where he improved the transformer profiler, added DML execution support, and resolved platform-specific build and dependency issues—and to Windows Machine Learning tools, enhancing full-stack features in React/TypeScript and Python. He focuses on making ML tooling more robust and interoperable, tackling nuanced compatibility problems like opset mismatches and model-loading edge cases. Colleagues rely on him for pragmatic fixes that bridge low-level runtime concerns and user-facing dashboard functionality, reflecting both systems and frontend strengths. An often-overlooked strength is his attention to cross-platform build subtleties, such as CMake tweaks and DLL dependencies, which keep complex ML pipelines working across environments.
10 years of coding experience
Github Skills (14)
machine-learning10
deeplearning-ai10
typescript10
typescripts10
deep-learning10
tensorflow10
onnx10
python10
hardware-acceleration10
react10
typescript-types10
windows9
pytorch9
cmake8
Programming languages (3)
Github contributions (5)
microsoft/Windows-Machine-Learning
Sep 2021 - Apr 2022
Samples and Tools for Windows ML.
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:49 reviews, 37 commits, 9 PRs in 7 months
Contributions summary:John primarily focused on enhancing the WinMLDashboard tool, specifically adding functionality for user-specified input and output names for TensorFlow models. They modified both the frontend (using React and TypeScript) and backend (Python) components of the dashboard. The user also addressed issues related to ONNX version compatibility, adding support for opset v14 and providing warnings for unmatched versions.
machine-learningwindowsaimldeep-learning
microsoft/onnxruntime
Oct 2021 - Mar 2022
ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
Role in this project:ML Engineer
ML EngineerContributions:7 reviews, 13 commits, 12 PRs in 5 months
Contributions summary:John primarily contributed to the ONNX Runtime project by implementing and improving the transformer profiler tool. Their work included adding support for the DmlExecutionProvider, fixing bugs within the profiler's Python scripts, and improving the tool's functionality. The user also addressed platform-specific issues, such as modifying CMake conditions and incorporating necessary dependencies like `api-ms-win-core-com-l1-1-0.dll`, further enhancing the tool's capabilities and compatibility. Additionally, the user worked on adding and reverting model loading from buffer, indicating a focus on model loading and deployment.
machine-learningonnxruntimedeep-learningonnxneural-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.