Jack Zhou
高级算法工程师 at 蚂蚁金服
Chaoyang District, Beijing, China
Summary
🤩
Rockstar🎓
Top School
Top expert inAdvanced Chinese Natural Language Processing and Machine Learning Technologies
Jack Zhou is a senior algorithm engineer with 11 years of experience specializing in machine learning infrastructure, model engineering, and production deployment, currently based in Chaoyang District, Beijing and working at Ant Financial. He has deep hands-on experience contributing to core components of the PaddlePaddle ecosystem—implementing losses and ops in the framework, integrating and optimizing transformer models like BigBird in PaddleNLP, and enabling efficient C++ batch inference for vision models in PaddleX. His work spans both research-oriented model development and practical serving/benchmarking, including CUDA kernel implementation, multi-process training, and segmentation service visualization. A Harbin Institute of Technology alumnus with a strong track record at Baidu and Ant, he combines rigorous academic training with production-proven engineering and a knack for improving documentation and developer experience.
11 years of coding experience
硕士, Computer Science, 硕士, Computer Science at 哈工大
Chinese, English
Github Skills (40)
tokenize10
c-language10
python10
classification10
pretrain10
machine-learning10
word-embeddings10
transformer-models10
api-documentation10
deeplearning-ai10
tokenizer10
image-segmentation10
deep-learning10
segmentation10
natural-language-processing10
Programming languages (4)
Github contributions (5)
PaddlePaddle/PaddleNLP
Feb 2021 - Jan 2023
Easy-to-use and powerful LLM and SLM library with awesome model zoo.
Role in this project:
ML Engineer
ML EngineerContributions:303 reviews, 180 commits, 293 PRs in 1 year 11 months
Contributions summary:Jack primarily contributed to the development of the BigBird model, a transformer-based model. Their work included integrating pretraining and classifier scripts, addressing model bugs, and implementing multi-process training. The user also focused on the integration and improvement of the existing models, enhancing the existing functionalities by adding more detail to the documentation, and making code format changes.
extractionsemantic-similaritybertnlp-tasksfinetuning
PaddlePaddle/Paddle
Aug 2020 - Apr 2022
PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）
Role in this project:ML Engineer
ML EngineerContributions:59 reviews, 41 commits, 74 PRs in 1 year 8 months
Contributions summary:Jack implemented and debugged the negative log likelihood (nll) loss API and integrated it into the paddlepaddle framework. Their contributions included adding the API definition, updating demo code, and adding a CUDA kernel for nll loss. They also added isfinite v2 op and contributed to enhancing the reduce operations within the framework. Furthermore, they optimized error descriptions for the math directory and contributed to the implementation of elementwise operations.
pytorchpythonparalleldeep-learningpaddlepaddle
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.