Xuan-son Nguyen
Ingénieur Logiciel at Hugging Face
Greater Paris Metropolitan Region France
Summary
🤩
Rockstar🎓
Top SchoolXuan-son Nguyen is a software engineer based in the Greater Paris area with 12 years of experience building performant, low-level systems and on-device ML runtimes. Currently at Hugging Face, he maintains GGUF and integrations for llama.cpp and contributes optimizations that make large-language-model inference practical in constrained environments. His open-source work includes meaningful performance and portability improvements to whisper.cpp and the ggml tensor library—optimizing SIMD for WASM, adding LoRA adapter support, and fixing platform-specific issues. Comfortable across C++, NodeJS, and systems-level networking, he has migrated codebases from Python to C++ and shipped cross-platform SDKs and apps. A security-trained engineer (INSA) with hands-on cybersecurity and backend experience, he pairs a pragmatic systems mindset with a flair for low-level optimization that often yields disproportionate performance gains.
11 years of coding experience
3 years of employment as a software developer
Diplôme d'ingénieur Sécurité et Technologies informatiques, Diplôme d'ingénieur Sécurité et Technologies informatiques at l'INSA Centre Val de LoireComputer Science, Computer Science at Aix-Marseille UniversityComputer Science, Computer Science at Vietnam National University, Hanoi
Vietnamese, English, French
Stackoverflow
376reputation
33kreached
11answers
0questions
Github Skills (30)
c-language10
algebra10
machine-learning10
c1110
c1710
performance-optimization10
optimisation10
simd10
blazor-webassembly10
tensor10
webassembly10
cprogramming-language10
optimization10
wasm-pack9
rust-wasm9
Programming languages (18)
Github contributions (5)
ggml-org/ggml
Jul 2024 - Mar 2025
Tensor library for machine learning
Role in this project:
ML Engineer
ML EngineerContributions:1 PR, 5 comments, 3 issues in 8 months
Contributions summary:Xuan-son primarily contributed to the development and improvement of the `ggml` tensor library for machine learning. Their work included implementing new features for LoRA adapter support, optimizing SIMD for WASM, and fixing issues related to unique tensor names and platform compatibility. Key contributions involved refactoring LoRA adapter functionalities, enhancing conversion scripts, and optimizing the library for improved performance in various environments.
deep-learningllmmachine-learningtensor-libraryneural-network
ggerganov/whisper.cpp
Sep 2024 - Jan 2025
Port of OpenAI's Whisper model in C/C++
Role in this project:Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:1 comment in 4 months
Contributions summary:Xuan-son primarily focused on low-level optimizations and refactoring within the `ggml` library of the Whisper.cpp project. Their contributions include enforcing unique tensor names, adding features for handling LoRA adapters, and significantly improving performance for WASM builds by optimizing SIMD operations. The commits also involved fixing platform-specific issues, such as those related to `cpu_set_t` on Emscripten, and adding functionality to the `gguf-split` utility.
cppwhisperspeech-to-textinferencespeech-recognition
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.