Da Yan
New York, New York, United States
Summary
🤩
RockstarDa Yan is a Member of Technical Staff based in New York with eight years of experience specializing in GPU compilers and high-performance AI kernels. He focuses on squeezing maximum performance from DNN and HPC primitives like GEMM, convolution, and reductions, and has hands-on experience improving Triton’s code generation and A100-specific shared memory optimizations. Prior roles include contributing to Triton as an independent contractor for OpenAI and building fast AI kernels at Anthropic, reflecting deep industry trust in his low-level GPU expertise. He holds a PhD in Computer Science from HKUST and a BE in Electrical Engineering, combining rigorous academic training with production-grade compiler engineering. Beyond bug fixes and feature work, he has a knack for subtle pointer- and memory-related optimizations that yield outsized performance gains on real hardware.
9 years of coding experience
Github Skills (11)
triton10
code-generation10
gpu10
optimisation10
optimization10
compiler-development10
pointer9
cprogramming-language9
c-language9
cuda8
llvm7
Programming languages (4)
Github contributions (5)
triton-lang/triton
Dec 2020 - Jan 2023
Development repository for the Triton language and compiler
Role in this project:
Backend Developer
Backend DeveloperContributions:22 reviews, 156 commits, 64 PRs in 2 years 1 month
Contributions summary:Da primarily contributed to the development of the Triton language compiler. Their work focused on implementing and improving the code generation phase, particularly concerning unary operations such as increment and decrement for various data types. They addressed and fixed bugs related to pointer arithmetic and memory management, and added support for key features, like prefix and postfix operations, which enhanced the language's capabilities. Furthermore, they implemented performance improvements for the A100 GPU, optimizing the shared memory loads and stores.
compiler
daadaada/turingas
Sep 2019 - Jan 2022
Assembler for NVIDIA Volta and Turing GPUs
Contributions:48 commits, 2 PRs, 28 pushes in 2 years 4 months
assemblernvidia
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.