Hector Yee
Staff Software Engineer at Waymo
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolHector Yee is a Staff Software Engineer with 20+ years building large-scale machine learning, computer vision, and graphics systems, now leading research in public and environmental health and generative vision for geospatial and climate adaptation. He’s shipped core recommendation and perception systems at YouTube, architected Airbnb’s interpretable Aerosolve pricing models, and contributed to TensorStore pipelines for multi-dimensional scientific data. Comfortable from low-level C++ rendering and real-time video classification to Hadoop/Spark pipelines and TensorFlow MLOps, he combines research rigor with product delivery—often translating complex inference problems into practical, parallelizable engineering tasks. Notably, his work spans Emmy-winning personalized video recommendations to inverse rendering of satellite imagery, reflecting a rare blend of creative graphics, optimization, and applied ML at web and terra-scale.
20 years of coding experience
22 years of employment as a software developer
MS Architectural Science (Computer Graphics), MS Architectural Science (Computer Graphics) at Cornell University
Chinese
Github Skills (17)
data-pipelines10
python10
machine-learning10
scala10
trainings10
data-processing10
data-pipeline10
apache-beam10
modeling10
java9
numpy9
linear-regression9
classification9
javas9
gcp8
Programming languages (4)
Github contributions (5)
airbnb/aerosolve
May 2015 - Feb 2016
A machine learning package built for humans.
Role in this project:
ML Engineer
ML EngineerContributions:260 commits, 81 PRs, 144 pushes in 9 months
Contributions summary:Hector's initial commit sets up the core library with the initial implementation of the MaxoutModel and its associated WeightVector class. The user further established the foundation for a training directory, including implementing the LinearRankerTrainer for regression and classification tasks. Moreover, the user contributes code to the creation of examples from pixels in the image impressionism demo, and implements the training code, along with model loading and transformation, for the demonstration of the system.
machine-learning
google/tensorstore
Apr 2020 - May 2020
Library for reading and writing large multi-dimensional arrays.
Role in this project:Data Engineer & Software Engineer
Data Engineer & Software EngineerContributions:9 commits, 1 issue in 13 days
Contributions summary:Hector primarily contributed to the development of a Beam pipeline for processing large multi-dimensional arrays, focusing on functionalities like rechunking and computing percentiles. They implemented and refined Python-based Beam pipelines leveraging TensorStore for data handling. The contributions also include adding tests, refactoring the pipeline into reusable components, and expanding the pipeline's capabilities, demonstrating a focus on data processing and system design.
multidimensional-arrays
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.