Madhur Amilkanthwar
Compiler Engineer at NVIDIA
Pune, Maharashtra, India
Summary
🤩
Rockstar🎓
Top SchoolMadhur Amilkanthwar is a Compiler Engineer with 10 years of experience, currently optimizing NVIDIA’s Grace superchip and contributing to high-performance compiler backends. He has a strong track record across NVIDIA, AMD, and Intel working on PTX, ROCm/AMDGPU, and heterogeneous ML ASIC compiler backends, with deep expertise in instruction selection, scheduling, register allocation, and polyhedral optimizations from an MS at IIT Madras. An active LLVM contributor, he improves AArch64 backend performance—adding NEON intrinsics, refining loop vectorization legality, and tuning scatter overheads for Neoverse-V2—showing low-level performance craftsmanship. Known for blending research-led compiler techniques with production-grade engineering, he brings both correctness-focused fixes and practical optimizations that materially boost real-world workloads.
10 years of coding experience
8 years of employment as a software developer
Indian Institute of Technology MadrasBachelor's degree, Computer Science, 9.2/10, Bachelor's degree, Computer Science, 9.2/10 at Government College of Engineering, Aurangabad
English, Hindi, Marathi
Github Skills (16)
compiler-optimization10
performance-tuning10
assembly10
c-language10
cprogramming-language10
vectorization10
assemble10
assembler10
llvm10
codegen9
g9
code-generation9
code-generator9
machine-code9
float328
Programming languages (3)
Github contributions (5)
llvm/llvm-project
Mar 2020 - Apr 2025
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:106 reviews, 43 PRs, 40 pushes in 5 years 1 month
Contributions summary:Madhur primarily contributes to the AArch64 backend of the LLVM compiler project, focusing on optimizations and performance improvements. Their work includes generating optimized libcalls for floating-point truncations and extensions, as well as adding support for the `neon.abs` intrinsic. The user also increased scatter overhead on the Neoverse-V2 architecture to enhance the performance of s128 kernels, demonstrating an understanding of low-level performance tuning. Furthermore, they fixed a bug in the loop vectorization legality check, improving the overall correctness of the compiler.
compilerstechnologiesclangsubmittoolchain
madhur13490/llvm-project
Apr 2020 - Mar 2025
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies. Note: the repository does not accept github pull requests at this moment. Please submit your patches at http://reviews.llvm.org.
Contributions:6 PRs, 104 pushes, 45 branches in 4 years 11 months
llvm-ircompilerstechnologiesclangbazel
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.