Ping Yu
Staff Software Engineer at Google
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPing Yu is a Staff Software Engineer in Mountain View with 11 years of experience building high-performance ML runtimes and large-scale web systems. At Google he helped found and grow TensorFlow.js into a widely used WebML solution (500M+ CDN downloads, 18K stars) and currently leads development of a unified on-device runtime spanning TFLite/TFJS/TFLM to enable GenAI across mobile, web, and embedded targets. He brings deep hands-on expertise in model conversion, optimization, and execution across CPU/GPU/NPU, and has contributed core operations (pow, norm, squeeze) and audio-model tooling to flagship TensorFlow.js repositories. Earlier roles show a mix of systems, networking, and full-stack leadership—from kernel and device-driver optimizations to SaaS platform delivery—demonstrating an ability to move between low-level performance work and product-facing engineering. Collected across top-tier open-source projects and large advertising products, his track record blends technical depth with sustained project growth and cross-platform pragmatism.
11 years of coding experience
9 years of employment as a software developer
M.S. Computer Science, M.S. Computer Science at University of MarylandBachelor of Science (B.S.) Computer Science, Bachelor of Science (B.S.) Computer Science at Tsinghua University
Chinese
Github Skills (15)
webgl10
unit-testing10
tensorflowjs10
machine-learning10
nodejs10
javascript10
model-conversion10
trainings10
tensorflow10
python10
modeling10
linear-algebra9
back-end-development9
audio-processing9
gpu-programming9
Programming languages (8)
Github contributions (5)
tensorflow/tfjs-converter
Feb 2018 - Aug 2019
Convert TensorFlow SavedModel and Keras models to TensorFlow.js
Role in this project:
Back-end Developer
Back-end DeveloperContributions:1 release, 298 commits, 294 PRs in 1 year 6 months
Contributions summary:Ping primarily worked on the TensorFlow.js converter, implementing features to convert TensorFlow models to the appropriate format for the web. They developed scripts for converting models, including freezing graphs, extracting weights, and generating JSON files. Their contributions focused on making the converted models compatible with TensorFlow.js by addressing issues with specific operations and implementing functionalities like a prelu fusion.
keras-modelsjavascriptnodejstensorflowtensorflow-js
tensorflow/tfjs-models
Jul 2018 - Oct 2022
Pretrained models for TensorFlow.js
Role in this project:Back-end Developer
Back-end DeveloperContributions:242 reviews, 72 commits, 253 PRs in 4 years 3 months
Contributions summary:Ping's commits primarily focus on developing and improving audio command models within the tensorflow/tfjs-models repository. They added utilities for audio data preprocessing and node.js model implementation, including training CLIs. Furthermore, the user addressed review comments and changed the node dependencies to use the CPU backend. These changes point towards the user focusing on backend and model-related development.
javascriptmachine-learningpretrained-modelstensorflowtensorflow-js
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.