Rebecca Qian
Co-Founder And CTO at Patronus AI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolRebecca Qian is a Co-Founder and CTO based in San Francisco with a decade of experience building production ML systems and embodied agent infrastructure. She led research and engineering at Facebook AI/FAIR, where she released FairBERTa and a demographic-perturbation model, managed semantic parsing for robotic assistants, and built human-in-the-loop pipelines to improve embodied agents. Her contributions span both foundational NLP fairness work and practical tooling—such as back-end and training improvements to the FacebookResearch FAIRO embodied-agent platform. At Patronus AI she now combines that research pedigree with product and engineering leadership to scale AI oversight, environment design, and reinforcement-learning-driven systems. Rebecca’s background in statistical pattern mining and real-world infra+ML deployments gives her a rare fluency in turning cutting-edge research into reliable, scalable products.
9 years of coding experience
4 years of employment as a software developer
Bachelor of Arts (B.S.) Honors Computer Science, Bachelor of Arts (B.S.) Honors Computer Science at University of Chicago
English, Chinese, Spanish
Stackoverflow
11reputation
491reached
0answers
1question
Github Skills (16)
transformers10
pytorch10
nlp10
machine-learning10
python10
bert10
natural-language-processing9
modeling9
trainings9
tokenizer8
beam-search8
tokenize8
deep-learning7
mongodb6
nosql6
Programming languages (6)
Github contributions (5)
facebookresearch/fairo
Dec 2020 - Aug 2021
A modular embodied agent architecture and platform for building embodied agents
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:110 reviews, 279 commits, 159 PRs in 8 months
Contributions summary:Rebecca updated and expanded several Python notebooks related to the training of the semantic parser, including documentation and the addition of new sections. The user made modifications to the training and evaluation process of the model, adding features such as fixed value head and supporting both string and array span definitions. These changes focused on improving the model's performance and expanding the parser's capabilities in the Droidlet project.
agentagentsembodied-agentmodulararchitecture
gurlcode/gurl
Aug 2016 - Mar 2017
Contributions:19 pushes, 2 branches in 7 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.