Ethan He
Member Of Technical Staff at xAI
California, United States
Summary
🤩
Rockstar🎓
Top SchoolEthan He is a Senior Deep Learning Engineer based in California with a decade of experience building and open-sourcing LLMs and generative AI systems. He has held research and engineering roles at NVIDIA, Facebook AI, and as a founding MLE at Headroom (acquired by Upwork), delivering production multimodal features like video summarization, real-time emotion/gaze/pose pipelines, and WebGL-based super-resolution. His research pedigree includes a CMU master’s in computer vision and a strong open-science footprint—several highly cited papers (≈6k citations) and active repositories such as an ICCV'17 channel-pruning implementation and U-Net work for biomedical segmentation. Ethan bridges research and production, routinely optimizing model architectures and integrating them into scalable Kubernetes deployments. Notably, he sustains a sizable developer following on GitHub and contributes practical model optimizations that accelerate very deep networks in real-world applications.
10 years of coding experience
5 years of employment as a software developer
Bachelor’s Degree, Computer Science, Bachelor’s Degree, Computer Science at Xi'an Jiaotong UniversityMaster's degree, Computer Vision, 4.0, Master's degree, Computer Vision, 4.0 at Carnegie Mellon University
English, Chinese
Github Skills (19)
pytorch10
convolutional-neural-networks10
python10
net10
model-compression10
keras10
image-segmentation10
deep-learning10
segmentation10
tensorflow10
deep-neural-networks10
neural-network10
image-recognition9
acceleration9
data-preprocessing9
Programming languages (8)
Github contributions (5)
ethanhe42/channel-pruning
Aug 2017 - Feb 2022
Channel Pruning for Accelerating Very Deep Neural Networks (ICCV'17)
Role in this project:
ML Engineer
ML EngineerContributions:4 releases, 54 commits, 1 PR in 4 years 7 months
Contributions summary:Ethan primarily contributed to the `net.py` file, which suggests involvement in the core neural network model implementation. Their changes focus on channel pruning techniques, including spatial and channel decomposition, as well as integration with other components like `builder.py` and `train.py`. The user also updated configurations and wrapper functions, implying active engagement in model optimization and integration within the deep learning pipeline.
mobilenetimage-recognitionaccelerationimage-classificationiccv
ethanhe42/u-net
Jun 2016 - Feb 2022
U-Net: Convolutional Networks for Biomedical Image Segmentation
Role in this project:ML Engineer
ML EngineerContributions:19 commits, 2 PRs, 5 pushes in 5 years 8 months
Contributions summary:Ethan primarily focused on developing and refining a U-Net model for biomedical image segmentation. Their commits demonstrate iterative improvements to the model architecture, including modifications to the convolutional layers, upsampling methods, and loss functions. The changes involve experimenting with techniques like dropout, different loss functions (e.g., quadloss), and adjustments to the model's depth and size. The user also made changes related to data preprocessing and augmentation to improve model performance.
pythonsemantic-segmentationdeep-learningimage-segmentationcomputer-vision
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.