Shanmugam Ramasamy
Senior Deep Learning Algorithms Engineer at NVIDIA
California, United States
Summary
🤩
Rockstar🎓
Top SchoolShanmugam Ramasamy is a Senior Deep Learning Algorithms Engineer at NVIDIA with a strong foundation from a Master’s in Computer Engineering at Georgia Tech and over three years focused on deep learning systems. He specializes in scaling and hardening training pipelines for large language and multimodal models, contributing to high-profile open-source projects like NVIDIA NeMo and the NeMo-Framework-Launcher. His contributions span test automation, validation of training metrics, dataset and batch handling fixes, and optimizations such as sequence parallel and O2/Apex support—work that helps move research-grade models toward reliable, production-ready training. Prior experience as a senior software engineer at Yahoo and an early data science internship at BMW reflect a pragmatic blend of product-minded engineering and research-driven rigor. Based in California, he pairs low-level training optimization skills with a keen eye for testability and reproducibility, often surfacing subtle failure modes through automated validation.
3 years of coding experience
6 years of employment as a software developer
Master of Science (M.S), Master of Science (M.S) at Georgia Institute of TechnologyAnna University, Chennai
English, Tamil
Github Skills (17)
pytorch10
pytest10
artificial-neural-networks10
machine-learning-models10
python10
testing10
machine-learning10
test-framework10
large-language-models10
tensorboard10
deep-learning10
bert10
neural-network10
test-automation10
generative-ai9
Programming languages (2)
Github contributions (5)
NVIDIA/NeMo-Framework-Launcher
Jul 2022 - Nov 2022
Provides end-to-end model development pipelines for LLMs and Multimodal models that can be launched on-prem or cloud-native.
Role in this project:
QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:2 reviews, 166 commits, 1 PR in 4 months
Contributions summary:Shanmugam's commits primarily focus on creating and refactoring testing frameworks for the Nemo framework-launcher repository. The code changes reveal the implementation of automated tests for training pipelines, specifically focusing on metrics related to loss and timing. The user has written tests to compare actual and expected training loss values, demonstrating expertise in assessing the performance and correctness of machine learning model training.
NVIDIA/NeMo
Sep 2022 - Dec 2022
A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech)
Role in this project:ML Engineer
ML EngineerContributions:107 reviews, 26 commits, 63 PRs in 2 months
Contributions summary:Shanmugam primarily focused on enhancing the NeMo framework for large language models. Their contributions involved bug fixes and improvements to the dataset and training processes, specifically related to validation batch sizes and supporting drop last batches. Additionally, they worked on integrating sequence parallel support and implementing O2 support with Apex pipeline functions, demonstrating efforts to optimize and scale the BERT model.
asrspeech-recognitionnatural-language-processingttsspeaker-diarization
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.