Vijay Thakkar
AI Research Scientist at Meta
Boston, Massachusetts, United States
Summary
🤩
Rockstar🎓
Top SchoolVijay Thakkar is an AI research scientist and compute architect with 9 years of deep experience in accelerated computing, performance engineering, and parallel algorithms, currently working on hardware-model co-design at Meta. He led the design of CUTLASS 3.x/4.x and the CuTe project at NVIDIA, collaborating across GPU architecture, compiler, and research teams to expose tensor-core features through CUDA C++ and PTX/MLIR toolchains. His background spans pre-silicon work for Hopper and Blackwell, sparse linear algebra prototyping, and practical HPC education—he co-ran Georgia Tech’s student cluster efforts and founded BU’s HPC club. Comfortable bridging low-level ISA/compiler details and high-level ML workloads, he combines a mechanic’s pragmatism (self-described “Systems and Performance Mechanic”) with rigorous academic training toward PhD-level research.
9 years of coding experience
7 years of employment as a software developer
PhD Computer Science, PhD Computer Science at Georgia Institute of TechnologyBachelor’s Degree Computer Engineering, Bachelor’s Degree Computer Engineering at Boston University
Github Skills (41)
pytorch10
memory-efficient10
python10
llama10
machine-learning10
inference10
linux-kernel10
amd10
llm10
nvidia10
deep-learning10
openai10
gpu10
cuda10
gpt10
Programming languages (5)
Github contributions (5)
thakkarV/cutlass
Mar 2020 - Jul 2024
CUDA Templates for SemiRing Linear Algebra Subroutines
Contributions:27 pushes, 14 branches in 4 years 5 months
cudalinear-algebrasemiringndarraygpu
thakkarV/lc-wavenet
Jul 2017 - Oct 2018
Contributions:109 commits, 124 pushes, 4 branches in 1 year 3 months
deep-learningdeepmindwavenettensorflow
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.