Baizhou Huang
PhD Candidate
Beijing, China
Summary
🤩
RockstarBaizhou Huang is a PhD candidate in Natural Language Processing at Peking University with five years of hands‑on experience in machine learning and model engineering. Based in Beijing, he blends academic research with practical open-source contributions, notably enhancing PaddleViT by implementing a one‑cycle learning rate scheduler, fixing pretrained weight loading, and adding MAE support. His work reveals a taste for production-ready research: improving training utilities and model reliability rather than only prototyping new architectures. Comfortable across vision and language modalities, he focuses on reproducible, testable code that bridges experiments to deployable components. Colleagues can expect a researcher-engineer who values rigorous testing and incremental system improvements that accelerate research adoption.
5 years of coding experience
Github Skills (10)
mle10
paddlepaddle10
computer-vision10
transformer10
machine-learning10
deeplearning-ai10
deep-learning10
python10
ml10
classification9
Programming languages (3)
Github contributions (5)
BR-IDL/PaddleViT
Nov 2021 - Dec 2021
:robot: PaddleViT: State-of-the-art Visual Transformer and MLP Models for PaddlePaddle 2.0+
Role in this project:
ML Engineer
ML EngineerContributions:4 reviews, 37 commits, 6 PRs in 1 month
Contributions summary:Baizhou contributed to the PaddleViT repository by implementing and testing a one-cycle learning rate scheduler within the image classification ConvMixer module. Their work included adding the scheduler to the `utils.py` file and creating a corresponding test file. The user also fixed issues with loading pretrained weights within the ViT model and added a MAE folder, which indicates involvement in advanced machine learning models.
mlppaddlepaddlecomputer-visionvittransformer
skpig/verl
Mar 2025 - Nov 2025
verl: Volcano Engine Reinforcement Learning for LLMs
Contributions:78 pushes, 8 branches in 8 months
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.