Geoffrey Irving
Chief Scientist at AI Security Institute
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolGeoffrey Irving is a Chief Scientist at the UK AI Security Institute with 25 years of experience bridging cutting‑edge ML research and production software engineering. He led scalable alignment and reflection efforts at DeepMind and OpenAI after engineering roles at Google Brain, bringing rare expertise in both AGI safety research and robust engineering practices. His background spans academic rigor (PhD from Stanford) and applied systems work from Pixar to D. E. Shaw, reflecting a talent for turning complex research into reliable code. An active open‑source contributor, he has improved foundational projects like NumPy and TensorFlow’s DeepMath (neural theorem proving), often focusing on portability and build stability—an underappreciated but critical part of shipping dependable ML systems. Based in London, he combines deep theoretical understanding with hands‑on fixes that make large codebases more maintainable and portable.
25 years of coding experience
16 years of employment as a software developer
California Institute of TechnologyPh.D. Computer Science, Ph.D. Computer Science at Stanford University
Stackoverflow
6,553reputation
277kreached
15answers
38questions
scala
top-5%
heap-memory
top-5%
linux
top-5%
jvm
top-5%
command-line
top-5%
Github Skills (28)
debug10
c-language10
python10
build-system10
numpy10
cprogramming-language10
fileheader9
jvm9
headers9
scala9
c119
linux9
command-line9
header-files9
c179
Programming languages (17)
Github contributions (5)
tensorflow/deepmath
Nov 2016 - Mar 2017
Experiments towards neural network theorem proving
Role in this project:
Back-end Developer
Back-end DeveloperContributions:35 commits, 15 PRs, 20 pushes in 4 months
Contributions summary:Geoffrey primarily focused on fixing compilation issues and adapting the code for different compiler environments, specifically targeting clang. They addressed issues related to the use of friend classes and replaced them with private tokens to improve portability. Furthermore, the user merged core kernel changes, fixed include guards across multiple header files, and corrected build processes, demonstrating a focus on code stability and build system maintenance within the project. These efforts suggest the user was involved in foundational code improvements and ensuring the project's overall build and compatibility.
theoremtheorem-provingdeep-learningneural-networksneural-network
numpy/numpy
Dec 2011 - Feb 2018
The fundamental package for scientific computing with Python.
Role in this project:Back-end Developer
Back-end DeveloperContributions:5 commits, 1 PR, 8 comments in 6 years 3 months
Contributions summary:Geoffrey primarily contributed to bug fixes and code improvements within the NumPy codebase. They addressed issues related to user-defined type resolution within the ufunc functionality, ensuring proper handling of custom data types. They also made modifications, including reverting a check for a null pointer, demonstrating an understanding of the C API and code stability considerations. Furthermore, the user implemented a code change that improved the usability of `flatnonzero` function.
lapackpythonmpindarrayconvolution
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.