Shraiysh V
Deep Learning Compiler Engineer at NVIDIA
United States
Summary
🤩
Rockstar🎓
Top SchoolShraiysh V is a Deep Learning Compiler Engineer with 8 years of systems and compiler experience, currently working at NVIDIA on performance-critical ML runtimes. He has a strong background in GPU compiler and frameworks from roles at AMD, Apple, and as an open-source contributor to the LLVM ecosystem, and he has contributed concrete improvements to TensorFlow by integrating cuDNN's new frontend API for Flash Attention. Trained at IIT Hyderabad and holding a master's in CS from Rice, he blends research-oriented rigor with production-facing engineering across compilers, CUDA/XLA, and ML primitives. Notably, his work focuses on squeezing low-level hardware primitives for high-throughput attention kernels—a niche that sits at the intersection of compiler design and deep learning performance engineering.
8 years of coding experience
3 years of employment as a software developer
Bachelor of Technology (Hons) Computer Science, Bachelor of Technology (Hons) Computer Science at Indian Institute of Technology HyderabadMaster's degree Computer Science, Master's degree Computer Science at Rice University
Stackoverflow
106reputation
24kreached
8answers
1question
Github Skills (17)
machine-learning10
deep-learning10
cudnn10
tensorflow9
cuda9
deep-neural-networks8
neural-network8
python7
firebase6
android-studio6
android6
google-cloud-firestore6
firebase-realtime-database6
regex6
java6
Programming languages (11)
Github contributions (5)
tensorflow/tensorflow
May 2020 - May 2020
An Open Source Machine Learning Framework for Everyone
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 2 commits, 4 PRs in 1 day
Contributions summary:Shraiysh made several contributions related to integrating cuDNN's new frontend API for Flash Attention within the TensorFlow framework. Their work included adding support for the new API and ensuring it replaced the existing implementation. The user's contributions were focused on utilizing cuDNN for optimized performance, specifically around flash attention capabilities. These changes involved code modifications within the XLA stream executor, particularly in the CUDA cuDNN-related files.
pythondata-sciencedeep-learningmlmachine-learning
shraiysh/xla
Feb 2024 - Apr 2025
A machine learning compiler for GPUs, CPUs, and ML accelerators
Contributions:490 pushes, 129 branches in 1 year 1 month
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.