Chen Peng-jen - Software Engineer at Facebook

Chen Peng-jen

Software Engineer at Facebook

New York, New York, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Chen Peng-jen is a software engineer with a decade of experience building large-scale systems, currently contributing at Facebook from New York. He has a strong foundation from National Taiwan University (BS and MS in Computer Science) and prior experience developing modular front-end controls for Microsoft’s web-based mobile Office suite. At Facebook he has applied back-end expertise to an influential open-source project, fairseq, improving multilingual translation, multi-GPU training reliability, and transformer modelling with practical bug fixes and added unit tests. Known for bridging research-grade ML toolkits and production engineering, he combines a deep academic background with hands-on fixes that improve inference consistency and model deployment.

10 years of coding experience

2 years of employment as a software developer

Bachelor's degree, Computer Science, 4.0, Bachelor's degree, Computer Science, 4.0 at National Taiwan University

jiangcui junior high school

Senior high school, Senior high school at Taipei Municipal Jianguo High School

English, Chinese

Github Skills (6)

transformer-models10

machine-translation10

pytorch10

artificial-intelligence10

python10

nlp9

Programming languages (3)

JavaScriptJupyter NotebookPython

Github contributions (5)

facebookresearch/fairseq

Oct 2018 - Oct 2022

Facebook AI Research Sequence-to-Sequence Toolkit written in Python.

Role in this project:

Back-end Developer

Contributions:34 commits, 5 PRs, 2 pushes in 4 years

Contributions summary:Chen's contributions focused on improving the `fairseq` toolkit, primarily addressing issues related to multilingual translation and model updates. They fixed bugs in the multilingual translation task related to multi-GPU training and to-many scenarios. The user also made enhancements to the transformer model, including making learned positional embedding optional. Furthermore, they addressed inference parameter inconsistencies and added unit tests.

pytorchnlpsequencepythontransformer-architecture

pipibjc/pytorch

Sep 2020 - Sep 2020

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Contributions:2 pushes, 1 branch in 1 day

pythongpu-accelerationdeep-learninggpuacceleration

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.

Request Free Trial