Margaret Li
Visiting Researcher at Meta
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolMargaret Li is a Machine Learning and NLP researcher-engineer with 11 years of experience, currently a Visiting Researcher at Meta and a PhD candidate in Computer Science at the University of Washington. Her work blends applied research and engineering—previous roles at Facebook AI Research, Grid AI/PyTorch Lightning, and internships at Meta and Facebook demonstrate depth in dialogue systems, sequence-to-sequence modeling, and RL. She contributes to high-profile open-source projects like ParlAI, adding machine-translation tasks and Fairseq integration that strengthen multilingual evaluation pipelines. Comfortable moving models from research prototypes toward reproducible training and validation setups, she pairs strong academic foundations from UPenn with hands-on system-building chops. An uncommon strength is her sustained engagement in teaching and mentorship across university and enrichment programs, reflecting an aptitude for translating complex ideas into practice.
11 years of coding experience
4 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science (ML, NLP), Doctor of Philosophy - PhD, Computer Science (ML, NLP) at University of WashingtonBachelor of Engineering (B.E.), Computer Science, Bachelor of Engineering (B.E.), Computer Science at University of Pennsylvania
English, Chinese
Github Skills (8)
sequence-to-sequence10
machine-translation10
nlp10
python10
fairseq10
natural-language-processing10
pytorch9
data-engineering8
Programming languages (5)
Github contributions (5)
facebookresearch/ParlAI
Sep 2017 - Feb 2021
A framework for training and evaluating AI models on a variety of openly available dialogue datasets.
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 9 commits, 46 PRs in 3 years 5 months
Contributions summary:Margaret made several contributions to the `parlai` repository, focusing on the integration and expansion of machine translation capabilities. These changes include adding a new WMT English-to-German task with associated build and agent implementations. The user also incorporated a FairseqAgent, indicating a focus on leveraging fairseq for sequence-to-sequence tasks. Further improvements included adding a validation set to the WMT data and minor fixes.
nlpdeep-learningdatasetmachine-learningtraining
hadasah/hadasah.github.io
Dec 2019 - Oct 2024
A Bootstrap 4 resume/CV theme created by Start Bootstrap
Contributions:15 pushes, 1 branch in 4 years 10 months
bootstrap4cv-themebootstrap5start-bootstrapcss
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.