Varun Gangal
Research Scientist at Patronus AI
Jersey City, New Jersey, United States
Summary
👤
Senior🎓
Top SchoolVarun Gangal is a research scientist with 11 years of experience specializing in LLM evaluation, model behavior and alignment, currently researching at Patronus AI after roles at Amazon AGI and ASAPP. He combines deep academic roots—a CMU PhD focused on generative NLG, data augmentation and evaluation benchmarks—with hands-on production work building post-trained, tool-enabled LLM agents for enterprise customer support. Varun has contributed to community tooling such as the widely used NL-Augmenter (implementing coreference-swap and speech conversion transforms) and helped create the GEM benchmark for standardized NLG evaluation. His publications span practical defenses against hallucination, block-sparse inference scaling, and dialog post-training, reflecting a strong bridge between research and deployable systems. Based in Jersey City, he brings both rigorous experimental methodology and pragmatic engineering to alignment and evaluation challenges in generative AI. An uncommon strength is his track record of turning lightweight data-augmentation ideas from CMU into reusable tools and benchmark components used by the broader NLP community.
11 years of coding experience
3 years of employment as a software developer
Indian Institute of Technology MadrasDoctor of Philosophy (Ph.D.) Computer Science, Doctor of Philosophy (Ph.D.) Computer Science at Carnegie Mellon University
English, Hindi, Marathi
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (6)
text-manipulation10
machine-learning10
nlp10
data-augmentation10
python10
speech-recognition7
Programming languages (3)
Github contributions (5)
GEM-benchmark/NL-Augmenter
Feb 2021 - Jul 2021
NL-Augmenter 🦎 → 🐍 A Collaborative Repository of Natural Language Transformations
Role in this project:
ML Engineer
ML EngineerContributions:35 reviews, 16 commits, 4 PRs in 5 months
Contributions summary:Varun primarily contributed to the development of natural language processing (NLP) transformations for the `nl-augmenter` repository. Their work involved implementing a coreference swapping perturbation, which included the use of the `spacy` and `neuralcoref` libraries. The user also integrated and expanded upon existing transformation logic and introduced new interfaces within the project. Additionally, they incorporated speech conversion capabilities using libraries such as `gtts` and `speech_recognition`.
nlpinformation-theorysentencedeep-learningmachine-learning
vgtomahawk/vgtomahawk.github.io
Aug 2017 - Feb 2025
Contributions:148 pushes, 1 branch in 7 years 6 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.