Ian Wright, PhD

I'm a Machine Learning Engineer and Data Scientist.

Currently: stealth mode. Previously: leading a team at GitHub applying ML, including LLMs, to help secure the world's code. Prior to that: Director of Data Science at Semmle (acquired by GitHub). I specialize in translating developments in academic ML to commercial products.

I have a PhD in Artificial Intelligence and a PhD in Economics. I've authored over 20 patents applying ML to diverse domains such as software security, e-commerce, software productivity, motion recognition and control systems.


Email: wrighti AT acm DOT org

Selected commercial products

In collaboration with colleagues:

LLMs for detecting security vulnerabilities in GitHub's Code Scanning product (public release soon).

Deep Learning for finding security vulnerabilities in software, a major new GitHub feature. See Code Scanning finds vulnerabilities using machine learning and Leveraging machine learning to find security vulnerabilities.

ML to produce software quality metrics in Semmle's LGTM (acquired by GitHub).

LiveMove motion recognition software for Nintendo and Sony (used in dozens of hit games, including Just Dance).

Real-time pitch recognition technology (with Charu Desodt) for Sony's SingStar franchise.

Neural network research

Most of my work is private. But I've shared some work on my public GitHub page.

I invented "∂B nets", which are a new type of neural network, trainable by backpropagation, that "harden" to discrete, boolean-valued functions without loss of accuracy (unlike existing binarisation approaches). The hardened net is compact (1-bit weights) and interpretable.
- GitHub repo: db-nets
- Lossless hardening with ∂𝔹 nets. I. Wright. In "Differentiable Almost Everything: Differentiable Relaxations, Algorithms, Operators, and Simulators", ICML 2023 Workshop, Honolulu, 2023.


As a fun exercise, and when I get the time, I'm developing a C++ implementation of a generic Datalog system. My goal is to maximise the amount of computation that is performed at compile-time.
- GitHub repo: datalog-cpp.

Selected machine learning papers

Lossless hardening with ∂𝔹 nets. I. Wright. In "Differentiable Almost Everything: Differentiable Relaxations, Algorithms, Operators, and Simulators", ICML 2023 Workshop, Honolulu, 2023.

The standard coder: a machine learning approach to measuring the effort required to produce source code change. I. Wright and A. Ziegler.  Proceedings of the 7th International Workshop on Realising Artificial Intelligence Synergies in Software Engineering, RAISE 2019, Montreal.

Measuring software development productivity: a machine learning approach. J. Helie, I. Wright, A. Ziegler. Paper presented at the ‘Machine Learning for Programming’ Workshop affiliated with the 2018 Federated Logic Conference (FLoC).

Reinforcement learning and animat emotions. I. Wright. From Animals to Animats 4: Proceedings of the Fourth International Conference on Simulation of Adaptive Behavior 1996, Cambridge, MA: The MIT Press/Bradford Books.

Selected economics papers

Marx's transformation problem and Pasinetti's vertically integrated subsystems. I. Wright. Cambridge Journal of Economics, Volume 43, Issue 1, January 2019, Pages 169–186.

A category-mistake in the classical labour theory of value. I. Wright. Erasmus Journal for Philosophy and Economics, Volume 7, Issue 1, Spring 2014, pp. 27-55. 

Implicit microfoundations for macroeconomics. I. Wright. Economics: The Open-Access, Open-Assessment E-Journal, Special Issue, "Reconstructing Macroeconomics" edited by Masanao Aoki and Hiroshi Yoshikawa, Vol. 3, 2009-19.

Selected philosophy of mind papers

My interest in AI naturally leads me to occasionally write on the philosophy of mind.

Loop-closing semantics. I. Wright. In: Wyatt, J.L., Petters D.D., Hogg, D.C. (Eds.) From animals to robots and back: reflections on the hard problems in the study of cognition, 2014. Cognitive Systems Monographs: Springer, Cham, pp. 219--253.

 Towards a design-based analysis of emotional episodes Wright, I., Sloman, A. & Beaudoin, L. P.. Philosophy Psychiatry Psychology, 3, 1996, pp. 101-126. Reprinted in “Artificial Intelligence: Critical Concepts in Cognitive Science”, Vol 4, Routledge, 2000.