Hugh Perkins
Member Of Technical Staff at Genesis AI
United States
Summary
👤
Senior🎓
Top SchoolHugh Perkins is a seasoned machine learning and NLP engineer with 16 years of experience building research and production systems, currently a Member of Technical Staff at Genesis AI. He has a strong open-source pedigree—authoring early PyTorch work, porting Torch to OpenCL (cltorch), creating DeepCL benchmarking tools, and building Coriander to compile CUDA for OpenCL devices. At ASAPP he bridged NLP research and product engineering, shipping scalable, secure systems and publishing work on dialog intent induction at EMNLP 2019. His contributions span low-level GPU backends (clBLAS, cutorch) to higher-level language models (char-rnn), reflecting rare expertise across compilers, performance engineering, and applied NLP. Fluent in Mandarin and with advanced CS training from Tsinghua, he combines deep systems chops with practical ML research.
16 years of coding experience
17 years of employment as a software developer
Bachelor’s Degree Natural Sciences, Bachelor’s Degree Natural Sciences at University of CambridgeMaster’s Degree Computer Science, Master’s Degree Computer Science at Tsinghua UniversityChinese (Mandarin), Chinese (Mandarin) at Shenzhen University
English, mandarin / putonghua, French
Stackoverflow
8,243reputation
5.4mreached
105answers
50questions
tensorflow
top-5%
cuda
top-5%
nvidia
top-5%
keras
top-5%
python
top-5%
gpu
top-5%
Github Skills (50)
opencl10
pytorch10
benchmark10
convolutional-neural-networks10
c-language10
game-engine10
spring10
python10
operation10
tensorrt10
mathematics10
gpu-programming10
torch-hub10
machine-learning10
benchmarking10
Programming languages (24)
Github contributions (5)
hughperkins/tf-coriander
Oct 2016 - Feb 2018
OpenCL 1.2 implementation for Tensorflow
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:21 releases, 646 commits, 9 PRs in 1 year 4 months
Contributions summary:Hugh contributed to the implementation of an OpenCL 1.2 implementation for Tensorflow. They focused on creating the `cl_platform.cc` and `cl_platform.h` files, which likely involved defining the OpenCL platform within the TensorFlow environment. Subsequent commits involved adding dependencies on EasyCL, Clew, and CUDA-on-CL, suggesting an effort to integrate OpenCL kernels with the existing CUDA infrastructure. They also made modifications to enable building the software, and some testing functionality.
cudanvidiagpuintelmac
hughperkins/coriander
Jun 2016 - Feb 2021
Build NVIDIA® CUDA™ code for OpenCL™ 1.2 devices
Role in this project:Back-end Developer
Back-end DeveloperContributions:12 releases, 2019 commits, 17 PRs in 4 years 9 months
Contributions summary:Hugh's commits primarily focus on removing build warnings within the CUDA code for OpenCL 1.2 devices. These efforts involve modifications to the source code across multiple files, including changes to CUDA event synchronizations and function declarations. The user also contributed to debugging and improving code compilation and error handling, indicating a focus on code quality and build processes for the CUDA code.
cudaamdnvidiagpuintel
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.