Ann Lee
Research Scientist at Meta
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolAnn Lee is a Research Scientist at Meta (Facebook AI Research) with eight years of experience specializing in speech recognition and audio pre-training. A PhD graduate from MIT's Spoken Language Systems Group, she combines deep academic expertise in machine learning and pattern recognition with hands-on engineering—contributing to high-profile open-source work such as improving audio pipelines in fairseq and integrating datasets like TIMIT into wav2vec-U. Her background spans practical systems research from unsupervised pronunciation error discovery to graph-based knowledge approaches at Microsoft, reflecting a knack for turning complex linguistic problems into robust, efficient code. Based in New York, she is known for refactoring core components and improving reliability in production ML toolkits, demonstrating both research rigor and production-minded engineering.
8 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy (PhD) Computer Science, Doctor of Philosophy (PhD) Computer Science at Massachusetts Institute of TechnologyBachelor of Science (BS) Electrical Engineering, Bachelor of Science (BS) Electrical Engineering at National Taiwan University
English, Chinese, Japanese
Github Skills (8)
pytorch10
machine-learning10
speech-recognition10
audio-processing10
python10
fairseq10
artificial-intelligence9
data-engineering8
Programming languages (3)
Github contributions (5)
facebookresearch/fairseq
Jan 2021 - Dec 2022
Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
Role in this project:
ML Engineer
ML EngineerContributions:6 reviews, 17 commits, 1 PR in 1 year 11 months
Contributions summary:Ann primarily contributed to fixing bugs and improving the efficiency of audio pre-training and speech recognition tasks within the fairseq framework. They addressed issues related to data loading, label alignment, and inference processes. The user also added support for new datasets, specifically TIMIT, integrating it within the wav2vec-U pipeline. Furthermore, they refactored code related to clustering and unit tests to improve the reliability of the model.
pytorchnlpsequencepythontransformer-architecture
wnhsu/FactorizedHierarchicalVAE
Sep 2017 - Jan 2018
Contributions:4 commits, 1 push, 2 comments in 4 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.