Nitish Keskar

Member Of Technical Staff at OpenAI

Palo Alto, California, United States
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts

Summary

🤩
Rockstar
🎓
Top School
Nitish Keskar is a Member of Technical Staff at OpenAI with a decade of experience building and fine-tuning large language models. Previously a Principal Research Scientist at Salesforce and a PhD candidate at Northwestern, he blends deep academic training with production-first ML engineering. He has contributed to high-profile open-source work on controllable transformer generation (salesforce/ctrl), including cross-framework conversions and sampling improvements that improve reliability in generation. Based in Palo Alto, he focuses on bridging research prototypes and scalable model implementations. Colleagues know him for pragmatic bug fixes that unlock model behaviors and for making complex sampling logic robust in real-world codebases.
code11 years of coding experience
job5 years of employment as a software developer
bookBachelor of Technology (BTech), Bachelor of Technology (BTech) at Veermata Jijabai Technological Institute
bookMaster's degree, Master's degree at Northwestern University
github-logo-circle

Github Skills (9)

transformer-models10
pytorch10
machine-learning10
nlp10
model-conversion10
python10
natural-language-processing10
language-modeling10
tensorflow9

Programming languages (4)

C++Jupyter NotebookMATLABPython

Github contributions (5)

github-logo-circle
salesforce/ctrl

Sep 2019 - Sep 2020

Conditional Transformer Language Model for Controllable Generation
Role in this project:
userML Engineer
Contributions:35 commits, 3 PRs, 36 pushes in 1 year
Contributions summary:Nitish primarily contributes to the development and improvement of a conditional transformer language model. Their commits involve bug fixes related to model generation logic, including top-k and nucleus sampling. They also implemented features like a "topn" flag for printing top-n candidates and conversions of the TensorFlow model to PyTorch, demonstrating their proficiency in model fine-tuning and cross-framework compatibility. Python3 compatibility and bug fixes were also addressed.
language-model
Contributions:31 pushes, 1 branch in 4 years 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.
Request Free Trial