Animesh Jain
Software Engineer at Facebook
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolAnimesh Jain is a software engineer with nine years of experience specializing in deep learning compilers and PyTorch tooling, currently working on TorchDynamo, AOTAutograd, and TorchInductor at Meta in the San Francisco Bay Area. He has a strong research and systems background—PhD-level training from University of Michigan—and prior applied science roles at AWS and internships at NVIDIA and Microsoft where he tackled memory and compilation challenges for DNNs. On GitHub he contributes to high-profile projects like functorch (JAX-like transforms for PyTorch) and Apache TVM, implementing operator mappings, memory-efficient fusion with NVFuser, and optimized INT8 kernels for Intel architectures. Colleagues rely on him for bridging compiler theory and production engineering: he combines low-level performance work (data layouts, int8 kernels, hardware-aware optimizations) with practical benchmarking and tooling. Notably, his past work reduced feature-map memory footprints twofold on GPU workloads, reflecting a pattern of impactful, efficiency-first contributions.
9 years of coding experience
9 years of employment as a software developer
Master’s Degree, Computer Science and Engineering, 4.0, Master’s Degree, Computer Science and Engineering, 4.0 at University of MichiganBachelor of Technology (B.Tech.), Electrical and Electronics Engineering, 9.37, Bachelor of Technology (B.Tech.), Electrical and Electronics Engineering, 9.37 at Institute of Technology, BHUHigh School, High School at Lucknow Public School
English, Hindi
Github Skills (21)
pytorch10
c-language10
back-end-development10
operation10
python10
tensorrt10
tensorflow10
convolution10
tensor10
cprogramming-language10
machine-learning9
performance-optimization9
compiler-compiler9
compiler9
cuda8
Programming languages (6)
Github contributions (5)
apache/tvm
Sep 2018 - Jan 2021
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
Backend Engineer
Backend EngineerContributions:74 reviews, 94 commits, 181 PRs in 2 years 4 months
Contributions summary:Animesh implemented and refined the INT8 convolution operator for Intel machines, focusing on the NCHWc data layout. They addressed implementation details such as integrating int8 kernels for convolution and improving the optimization of specific Intel architectures (e.g., Skylake). Their contributions included code changes to the compilation stack, fixing bugs, and documentation, indicating involvement in enhancing performance and usability of the deep learning compiler.
metalvulkancompilertensoropencl
pytorch/functorch
Oct 2021 - Aug 2022
functorch is JAX-like composable function transforms for PyTorch.
Role in this project:Back-end Developer
Back-end DeveloperContributions:18 reviews, 134 commits, 72 PRs in 10 months
Contributions summary:Animesh contributed to operator authoring within the functorch library, specifically focusing on mapping PyTorch operations to ExprHandles. Their work involved adding and testing various unary and binary operators. The user also implemented memory-efficient pointwise fusion using NVFuser and created benchmarking scripts for LightSeq patterns. Further contributions include adding recomputation for forward passes in backward passes.
pytorchgradientshessianscomposablejax
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.