Philippe Tillet

San Diego / San Francisco, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

Philippe Tillet is a Member of Technical Staff at OpenAI with 13 years of experience in high-performance ML systems, compiler engineering, and GPU kernel optimization, based in San Diego. He specializes in code generation, auto-tuning and numerical kernels—work informed by optimized GEMM kernel development at AMD/NVIDIA and performance-portable linear algebra from his academic research. An active open-source contributor, he has improved the Triton language and compiler (driver API simplification, LLVM backend fixes and new operators like cos/sin) and implemented and debugged the Nearest Centroid classifier in the Shōgun toolbox, spanning compiler backends to ML algorithms. His background includes doctoral studies at Harvard SEAS and a master’s in EECS from NCTU, plus internships that delivered practical advances in PTX tuning and BLAS integration. Known for a meticulous focus on numerical correctness and memory management, he pairs small, correctness-driven fixes with broader refactors that raise performance and maintainability.

14 years of coding experience

Github Skills (19)

c-language10

llvm10

machine-learning10

compiler-design10

language-design10

cuda10

c-programming-language10

data-science9

datastructures9

datastructure9

data-structure9

algorithm9

optimizations9

algorithms9

cmake9

Programming languages (5)

C++ShellCMLIRPython

Github contributions (5)

triton-lang/triton

Aug 2014 - Jan 2023

Development repository for the Triton language and compiler

Role in this project:

Back-end Developer

Contributions:3 releases, 983 reviews, 2115 commits in 8 years 6 months

Contributions summary:Philippe's commits primarily focused on the development and maintenance of the Triton language and compiler. They made significant contributions to the driver API, simplifying its reliance on driver::context, and fixed multiple bugs related to the LLVM backend, the code generation, and the interpreter. Moreover, the user added the support for new operators like `cos` and `sin`, demonstrating an ability to extend the language.

compilerprogramming-languagecode-generationtriton

shogun-toolbox/shogun

Apr 2012 - Apr 2012

Shōgun

Role in this project:

ML Engineer

Contributions:7 commits in 14 days

Contributions summary:Philippe primarily contributed to the Shōgun machine learning toolbox by implementing and refining the Nearest Centroid classifier. Their work involved adding the classifier's core functionality, integrating it with existing components, and addressing memory management concerns. Further improvements included refactoring the code for clarity and efficiency, along with fixing bugs in the dot product calculations.

cmakedata-sciencegunc-plus-plusmachine-learning

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.

Request Free Trial