Sunny Guha
Applied Scientist III at Amazon
Natick, Massachusetts, United States
Summary
👤
Senior🎓
Top SchoolSunny Guha is an Applied Scientist III with eight years of experience specializing in large-scale LLM architecture, pretraining, and efficient fine-tuning for tool use and API calling. At Amazon Alexa AI he scales and fine-tunes 100B-parameter models on H100 and Trainium using Nemo Megatron, and has productionized transformer recommender systems serving millions of FireTV users. His background includes deep learning and computer vision engineering at MathWorks (shipping a DNN pruning library in MATLAB) and prior research rigor from a PhD in theoretical physics, where he contributed multiple publications on string theory and CFTs. He combines strong systems engineering—distributed training, quantized GPU code generation, MLPerf benchmarking—with theoretical mathematical instincts, enabling practical solutions like personalized language model integration that improved ASR metrics. Based in Natick, MA, he bridges academia and production ML, often applying curriculum learning and instruction-tuning to balance model generalization and memorization. An unexpected strength is translating advanced theoretical physics problem-solving into scalable ML training strategies that drive measurable product impact.
8 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy (PhD) Theoretical Physics, Doctor of Philosophy (PhD) Theoretical Physics at Texas A&M UniversityMaster of Science (MSc) Physics, Master of Science (MSc) Physics at Birla Institute of Technology and Science, Pilani - Goa Campus
Github Skills (10)
machine-learning8
pytorch8
python8
tensorflow7
data-science7
deep-learning7
neural-network5
jax4
solver3
bootstrap2
Programming languages (2)
Github contributions (5)
sol0invictus/YT-codes
Oct 2021 - Jan 2022
Contributions:12 commits, 11 pushes, 1 branch in 2 months
sol0invictus/MAT-DL
Jan 2021 - Apr 2021
Contributions:13 commits, 14 pushes, 1 branch in 2 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.