Ahmed Taei
Principal Engineer at NVIDIA
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolAhmed Taei is a Principal Engineer with a decade of experience building ML compilers, runtimes, and systems-level tooling, currently driving architecture at NVIDIA from Seattle. He specializes in MLIR-based backends and lowering pipelines—contributing to prominent open-source projects like IREE and Torch-MLIR where he implemented MHLO→Linalg/LLVM lowering, dynamic-shape bufferization, and Bazel build support. His background spans research and production roles at Google, Facebook AI, Cruise and Modular, blending compiler theory, performance engineering and pragmatic build/devops work. Ahmed has extended domain-specific languages such as Halide to support richer generator parameters and has repeatedly pushed E2E performance for CPU and accelerator targets. Colleagues rely on him for solving difficult cross-cutting problems between compilers, programming languages and hardware-software interfaces. He pairs deep academic training in applied mathematics with hands-on systems craftsmanship that surfaces in high-impact OSS contributions.
10 years of coding experience
13 years of employment as a software developer
Master of Science - MS Applied Mathematics, Master of Science - MS Applied Mathematics at University of WashingtonB.S Computer Engineering, B.S Computer Engineering at Cairo University
Github Skills (15)
compiler-optimization10
compiler10
halide10
c-language10
lg10
compiler-compiler10
cprogramming-language10
mlr10
bazel10
cicd10
dsl10
llvm10
python9
unit-testing8
arm7
Programming languages (5)
Github contributions (5)
iree-org/iree
Jan 2020 - Nov 2021
A retargetable MLIR-based machine learning compiler and runtime toolkit.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:256 reviews, 152 commits, 206 PRs in 1 year 9 months
Contributions summary:Ahmed primarily contributed to the IREE (Intermediate Representation Execution Environment) compiler, specifically focusing on lowering high-level operations (MHLO) to the Linalg dialect and subsequently to LLVM. Their work involved implementing patterns to convert various MHLO operations, such as convolution and reshape, into Linalg named operations. They also worked on the bufferization of these operations and the conversion to a specific ABI to be consumed by LLVM's JIT and AOT backends. Further work involved inlining, memory allocation and handling of shape and push constant variables to support dynamic shapes.
mlirspirvvulkantensorflowcompiler
llvm/torch-mlir
Mar 2022 - Jan 2023
The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.
Role in this project:Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:30 reviews, 14 commits, 29 PRs in 9 months
Contributions summary:Ahmed primarily contributed to the build system and codebase improvements. Their work included adding Bazel build support, which is a critical infrastructure task. Further changes involved refactoring code and adding new components by creating/updating Python scripts and makefile modifications. The user demonstrated expertise in build processes.
pytorchmlirtorchcompilerecosystem
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.