David Watson
Senior Staff Engineer at Graphcore
Bristol, England, United Kingdom
Summary
👤
Senior🎓
Top SchoolDavid Watson is a Senior Staff Engineer with 13 years' experience who combines deep mathematical research (PhD, University of Exeter) with hands-on systems engineering for ML hardware and compilers. At Graphcore he is a key contributor to the PyTorch backend/compiler for custom silicon, bridging software and hardware teams by flagging design issues and producing widely used training materials. Previously at Hawk-Eye he built advanced human skeletal recognition and tracking systems using a mix of classical and machine learning techniques. An active contributor to the core PyTorch codebase, he focuses on robustness—bug fixes, C++ cleanups, and precise numerical behavior—showing a preference for correctness over flashy features. Based in Bristol, he advocates modern tooling and continuous learning, often creating internal tools to keep codebases current and maintainable.
13 years of coding experience
3 years of employment as a software developer
Master's degree, Mathematics, Master's degree, Mathematics at University of CambridgeBachelor's degree, Mathematics, First Class Honours, Bachelor's degree, Mathematics, First Class Honours at University of WarwickDoctor of Philosophy - PhD, Mathematics, Doctor of Philosophy - PhD, Mathematics at University of Exeter
Github Skills (11)
pytorch10
machine-learning10
c-language10
tensor10
cprogramming-language10
python9
autograd9
neural-network9
debug8
debugging8
gpu8
Programming languages (6)
Github contributions (5)
pytorch/pytorch
Jun 2022 - Oct 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
Back-end Developer
Back-end DeveloperContributions:13 reviews, 9 PRs, 50 comments in 1 year 4 months
Contributions summary:David primarily focused on bug fixes and improvements related to the core functionality of the PyTorch library. Their contributions involved addressing issues within the C++ code, specifically removing redundant elements and implementing size checks. They also added decomposition functionality for the 'normal' operation, and included tests for the interpolate nearest exact method. These changes suggest a focus on refining existing features and ensuring the library's stability and correctness.
pythongpu-accelerationdeep-learninggpunumpy
Anstow/configsettings
Apr 2015 - Dec 2021
Contributions:49 pushes, 2 branches in 6 years 9 months
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.