Akshay Subramaniam
Principal AI Research Engineer, DevTech at NVIDIA
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolAkshay Subramaniam is a Principal AI Research Engineer at NVIDIA with 11 years of experience at the intersection of AI research, high-performance computing, and scientific modeling. He specializes in scientific machine learning and efficient training of complex models on large-scale GPU systems, translating new algorithms into production-ready, performance-tuned implementations. His background includes scalable fluid dynamics simulators and GAN-based turbulence enrichment developed during a PhD at Stanford, and hands-on CUDA kernel optimizations for the widely used Kaldi ASR project, including TF32 tensor core support. Based in Palo Alto, he pairs deep academic training with proven experience optimizing code for leadership-class supercomputers and production GPUs, making him adept at both algorithmic innovation and low-level performance engineering.
10 years of coding experience
12 years of employment as a software developer
Indian Institute of Technology MadrasDoctor of Philosophy (PhD), Aerospace, Aeronautical and Astronautical Engineering, Doctor of Philosophy (PhD), Aerospace, Aeronautical and Astronautical Engineering at Stanford University
English, Hindi, Tamil, Telugu
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (11)
tensorrt10
cuda10
c-language10
tensor10
cublas10
cpp10
cprogramming-language10
cplus10
speech-recognition9
speech-to-text9
machine-learning8
Programming languages (10)
Github contributions (5)
kaldi-asr/kaldi
Oct 2019 - Jan 2021
kaldi-asr/kaldi is the official location of the Kaldi project.
Role in this project:
ML Engineer
ML EngineerContributions:6 PRs, 14 comments, 1 issue in 1 year 2 months
Contributions summary:Akshay contributed to improving CUDA kernels for matrix operations, specifically for row sums and trace calculations, which are fundamental components in deep learning. They optimized existing kernels for better performance, including adding support for TF32 tensor cores to accelerate computation. The user's work also includes fixing test failures and upgrading existing CUDA-related functionalities to move away from deprecated features.
cudakaldiasrspeech-to-textkaldi-asr
akshaysubr/WCHR-regent
Jan 2017 - Mar 2019
Regent implementation of the WCHR scheme
Contributions:180 commits, 4 PRs, 29 pushes in 2 years 2 months
schemereacttypescript
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.