Jonathan Shen
Principal AI ML Engineer at Upwork
California, United States
Summary
🤩
Rockstar🎓
Top SchoolJonathan Shen is a Principal AI/ML Engineer based in California with 13 years of software experience, currently leading the technical development of Uma, Upwork’s Mindful AI. He brings deep research-to-production expertise from Google Research, where he created Lingvo and co-authored Tacotron 2 and Non-Attentive Tacotron—work that helped shape modern TTS and sequence modeling across multiple Google products. Jonathan pairs rigorous academic training (MSc from CMU, BSc from UBC) with hands-on systems engineering in Go and large-scale data pipelines from his time at Inference.io and a stealth startup exploring LLMs. He has a track record of improving code quality and maintainability in major open-source projects (notably tensorflow/lingvo) and of turning cutting-edge research into usable, real-world systems. Known for bridging deep learning research and production engineering, he also has a history of impactful accessibility work, such as creating custom voice models for people with ALS.
13 years of coding experience
9 years of employment as a software developer
Bachelor of Science (BSc) Computer Science and Math, Bachelor of Science (BSc) Computer Science and Math at The University of British ColumbiaMaster of Science Computer Vision, Master of Science Computer Vision at Carnegie Mellon University
English, Chinese, Japanese
Github Skills (8)
tensorflow10
python10
code-formatting10
refactoring9
nlp8
machine-translation6
speech-to-text6
distribute5
Programming languages (11)
Github contributions (5)
tensorflow/lingvo
Sep 2018 - Aug 2021
Lingvo
Role in this project:
Software Engineer (Focus on Code Quality and Formatting)
Software Engineer (Focus on Code Quality and Formatting)Contributions:5 reviews, 11 commits, 9 PRs in 2 years 11 months
Contributions summary:Jonathan's commits primarily focus on code formatting and style improvements within the Lingvo project. This includes fixing import sorting issues, removing trailing spaces, and enforcing an 80-character line length limit. These changes were applied to multiple files, indicating a project-wide effort to improve code readability and maintainability. The edits span different areas of the project, suggesting the user is contributing to a common standard for the codebase.
asrtranslationctcspeech-recognitiontensorflow
hypetrainai/hypetrain
Jul 2018 - Apr 2023
Main repository for the hypetrain team.
Contributions:2 PRs, 253 pushes, 6 branches in 4 years 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.