From 2018 to 2025 I was a Senior UX Researcher at MathWorks, leading research across five or more product teams building AI and deep learning tools for MATLAB and Simulink.
The object of study was code. The usability of a function lives in its name, its arguments, the error it produces when you get it wrong, and the example that appears when you search. To see it failing, you watch an engineer try to get something working and narrate their expectations, then check at scale whether what you saw generalises.
The work
- Large scale global surveys on needs, satisfaction and behaviour across a diverse user base of scientists and engineers who came to programming for a reason other than programming.
- Usability studies of programming interfaces, feeding into features that shipped, including the trainnet training function, dlnetwork and deep learning metrics.
- Data-driven user profiles for AI, giving teams a shared vocabulary for who they were building for. Presented publicly as The Diverse Seven at UXPA Boston 2022.
- Building the practice. Coaching designers and engineers to run their own usability tests and surveys, with templates and review, which raised how much research the organisation could absorb.
Public documentation I contributed to
- End-to-end AI workflows
- Train a neural network with tabular data
- Retrain a neural network to classify new images
I gave a guest lecture on interaction design at the University of Cambridge during this period.