JiahuiYu
Research Scientist
Summary
🤩
RockstarResearch Scientist with 9 years of experience specializing in applied machine learning and model development, particularly in computer vision and efficient neural networks. Notable open-source contributions include refactoring and extending influential projects like Slimmable Networks (ICLR/ICCV) and DeepFill inpainting (CVPR/ICCV), demonstrating deep expertise in architecture design, training pipelines, and practical releases. Comfortable moving research ideas into usable code, from integrating contextual attention and gated convolutions to implementing AutoSlim variants and batch inpainting features. Practical focus on improving model usability and performance, with a track record of cleaning, modularizing, and shipping research code. Often works at the intersection of cutting-edge publications and production-ready implementations.
9 years of coding experience
Github Skills (17)
pytorch10
python10
generative-adversarial-network10
mask-rcnn10
tensorflow10
deep-neural-networks10
neural-network10
faster-rcnn10
efficientnet10
neural-architecture-search10
adaption9
adaptation9
computer-vision9
automations8
automator8
Programming languages (2)
Github contributions (5)
JiahuiYu/generative_inpainting
Jan 2018 - Jun 2020
DeepFill v1/v2 with Contextual Attention and Gated Convolution, CVPR 2018, and ICCV 2019 Oral
Role in this project:
ML Engineer
ML EngineerContributions:42 commits, 11 PRs, 44 pushes in 2 years 4 months
Contributions summary:JiahuiYu focused on refactoring and releasing code related to the deep inpainting model. They implemented changes to the inpaint model, including adjustments to the network architecture and integration of contextual attention mechanisms. Furthermore, they contributed to the codebase by supporting batch inpainting with shared masks and releasing a new version of the software, indicating a focus on improving model functionality and usability.
pytorchiccviccv-2019deep-learninggenerative-adversarial-network
JiahuiYu/slimmable_networks
Dec 2018 - Sep 2020
Slimmable Networks, AutoSlim, and Beyond, ICLR 2019, and ICCV 2019
Role in this project:ML Engineer
ML EngineerContributions:32 commits, 11 PRs, 35 pushes in 1 year 8 months
Contributions summary:JiahuiYu primarily contributed to the model development and training aspects of the slimmable networks project. This is evident through the refactoring and release of code, alongside the addition of new models (USNet, AutoSlim) and modifications to core model architectures (e.g., autoslim_resnet, autoslim_mnasnet). The changes span across various files including `train.py` indicating work on the training pipeline and the `models` directory reflecting deep involvement in the model design and implementation, showcasing their expertise in the project's core functionality.
pytorchiccviccv-2019iclrimagenet
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.