Fei Xia
Senior Staff Research Scientist, Tech Lead Manager at Google DeepMind
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolFei Xia is a Senior Research Scientist at Google DeepMind Robotics with 12 years of experience building foundation models for embodied agents that bridge high-level planning and low-level control. He holds a PhD from Stanford and a perfect GPA from Georgia Tech, and focuses on scalable, transferable simulation, long-horizon learning algorithms, and unified geometric-semantic environment representations for home robotics. His work spans both research and engineering—contributing backend and renderer integration to the iGibson simulator and implementing CompletionNet and PointNet variants for 3D perception and segmentation. At Google he has led efforts on robotics foundation models, translating simulation-scale ideas into systems that work in complex, unstructured real-world scenes. Fei’s background uniquely blends rigorous theory (robot learning and 3D vision) with production-oriented code and open-source contributions, reflecting a knack for shipping reproducible research. He is based in Mountain View and often surfaces less obvious strengths: deep systems-level contributions to simulators that enable long-horizon, transferrable robot learning.
12 years of coding experience
11 years of employment as a software developer
Stanford SoE UGVRElectrical and Computer Engineering, Electrical and Computer Engineering at Georgia Institute of TechnologyBachelor of Engineering (BE), Automation, Bachelor of Engineering (BE), Automation at Tsinghua UniversityDoctor of Philosophy (PhD), EE, Doctor of Philosophy (PhD), EE at Stanford University
English, Chinese
Github Skills (18)
pytorch10
convolutional-neural-networks10
python10
machine-learning10
deep-learning10
trainings10
neural-network10
computer-vision10
point-cloud10
modeling10
evaluation9
eval9
data-preprocessing9
cprogramming-language8
c-language8
Programming languages (12)
Github contributions (5)
StanfordVL/iGibson
Nov 2017 - May 2022
A Simulation Environment to train Robots in Large Realistic Interactive Scenes
Role in this project:
Back-end Developer
Back-end DeveloperContributions:10 releases, 4 reviews, 2236 commits in 4 years 6 months
Contributions summary:Fei primarily focused on merging branches, specifically related to the "icra_submission" branch. These merges included changes to the "mesh_renderer" and "envs/locomotor_env.py" files, suggesting involvement in the rendering and environment setup for the iGibson simulator. The commits indicate a focus on integrating various components within the simulation environment, and on setting up the top-down renderer.
roboticssimulationrobot-simulatorrealistic3d-scenes
StanfordVL/GibsonEnv
Jul 2017 - May 2021
Gibson Environments: Real-World Perception for Embodied Agents
Role in this project:ML Engineer
ML EngineerContributions:1 release, 580 commits, 37 PRs in 3 years 10 months
Contributions summary:Fei appears to be working on a computer vision project related to real-world perception for embodied agents. They implemented a CompletionNet, a deep learning model, along with supporting code in datasets.py and utils.py. The user's commits indicate a focus on building and testing the neural network architecture, including adding adaptive normalization and experimenting with perceptual loss functions within a deep learning framework.
agentsenvironmentsgibsonperception
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.