Richard Pang
Research Scientist at Meta
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolRichard Pang is a research scientist with nine years of experience bridging academic rigor and industry-scale AI research, currently at Meta FAIR where he focuses on large-model post-training work such as Llama 3.3 and LM coding/reasoning. His background includes visiting and intern roles at Google Brain/DeepMind and Google Research, plus a PhD from NYU's Courant Institute and a strong quantitative foundation from the University of Chicago. He contributes to open-source NLP tooling—e.g., implementing multiple-choice tasks in the widely used jiant toolkit—demonstrating both practical engineering and task-design instincts. Based in New York, he blends deep theoretical training with hands-on model development and evaluation, often working on the less-visible but crucial integration and evaluation pieces that make research reproducible and deployable.
9 years of coding experience
2 years of employment as a software developer
Doctor of Philosophy - PhD, Courant Institute of Mathematical Sciences, Doctor of Philosophy - PhD, Courant Institute of Mathematical Sciences at New York UniversityBachelor of Science - BS, Computer Science (Honors), Mathematics, Statistics, 3.852/4.000, General Honors, Departmental Honors, Bachelor of Science - BS, Computer Science (Honors), Mathematics, Statistics, 3.852/4.000, General Honors, Departmental Honors at University of Chicago
English, Chinese
Github Skills (5)
transformers10
nlp10
python10
bert9
multi-task-learning8
Programming languages (3)
Github contributions (5)
nyu-mll/jiant
Oct 2019 - Dec 2020
jiant is an nlp toolkit
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 18 commits, 5 PRs in 1 year 2 months
Contributions summary:Richard primarily contributed to the implementation of multiple-choice tasks related to the ARC and Mutual datasets within the `jiant` NLP toolkit. Their commits involved creating new task classes, defining data structures, and integrating these tasks with the existing framework. They also made changes to the retrieval, and evaluation scripts, and included the new tasks in various project files, demonstrating an understanding of the project's architecture.
nlptransformersmultitask-learningsentence-representationbert
nyu-mll/quality
Dec 2021 - Dec 2022
Contributions:65 commits, 61 pushes, 8 comments in 1 year
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.