Mourad Gouicem
Senior Software Engineer at NVIDIA
Caen, Normandy, France
Summary
đ€©
Rockstarđ
Top SchoolMourad Gouicem is a senior software engineer with nine years of experience specializing in high-performance numerical and deep learning libraries, currently based in Caen, Normandy and working at NVIDIA. He spent eight years at Intel where he contributed to MKL-DNN/oneDNN and math libraries for compilers, focusing on low-level optimizations, AVX512 kernels, Winograd convolutions and low-precision matrix multiplication. His PhD and postdoc work in numerical computation and discrete logarithms reflect a strong theoretical foundation that he applies to practical performance engineering. An active open-source contributor, his commits to prominent projects like oneDNN and gemmlowp show attention to platform-specific optimizations and kernel dispatch issues that materially improve runtime efficiency. Colleagues describe him as a researcher-turned-engineer who bridges rigorous algorithms with production-grade, CPU/GPU-optimized implementations.
9 years of coding experience
11 years of employment as a software developer
Doctor of Philosophy (PhD) Computer Science, Doctor of Philosophy (PhD) Computer Science at Pierre and Marie Curie University
French, English
Github Skills (22)
performance-monitor10
c-language10
avx10
matrix-multiplication10
fixed-point10
standard-library10
fixed10
deep-learning10
performance-analysis10
deep-neural-networks10
performance-optimization10
neural-network10
convolution10
c-library10
common-library10
Programming languages (5)
Github contributions (5)
uxlfoundation/oneDNN
Jun 2017 - Dec 2022
oneAPI Deep Neural Network Library (oneDNN)
Role in this project:
Back-end Developer
Back-end DeveloperContributions:725 reviews, 464 commits, 149 PRs in 5 years 7 months
Contributions summary:Mourad primarily contributed to improving the oneAPI Deep Neural Network Library (oneDNN). Their work involved fixing dispatching issues and adding kernels for the AVX512_common instruction set, specifically related to convolution operations and Winograd algorithms. The commits suggest a focus on optimizing performance and expanding the library's capabilities for various deep learning applications. They also added and modified code in core engine file.
bfloat16sse41avx512openmpopencl
google/gemmlowp
Oct 2016 - Nov 2016
Low-precision matrix multiplication
Role in this project:Backend Developer
Backend DeveloperContributions:16 commits, 3 comments in 1 month
Contributions summary:Mourad's commits primarily focus on low-precision matrix multiplication and related optimizations, as indicated by the repository's description. They made several changes to the code, including updates to fixed-point implementations and SSE4 optimizations within the internal headers, suggesting the user contributed to performance improvements and platform-specific enhancements. The commits also include adjustments to test files and code, indicating active participation in improving the library's functionality.
precisionmatrix-multiplicationmatrixmultiplication
Find and Hire Top DevelopersWeâve 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.