PhD Engineering researcher at Imperial College London (Shawki Saad Scholar), on the VIEW project, funded by Schmidt Sciences.
Focus: Uncertainty quantification for Edge AI. Put simply; how far can a measurement be trusted, and what is it worth to a decision?
A growing share of environmental sensing now happens on small devices, where sensors and TinyML models make predictions on the spot. My research measures how far each of those predictions can be trusted. The people who rely on them shape that work: their experience decides how uncertainty is measured and which mistakes matter most.
This way of working is inspired by:
- my kind supervisor Wouter Buytaert, whose expertise in socio-hydrology studies water and people as one system;
- the ever-inspiring James Manyika, whose career path led to his leadership of Research, Labs, Technology & Society at Google;
- Demis Hassabis, the whiz kid doing the unusual!
- Richard Hamming, of blessed memory, whose talk You and Your Research asks why we don't work on the important problems.
- ng-flood: Google flood forecasts for any point in Nigeria, by postcode area.
- ng-postcode: Nigeria's NIPOST digital postcode in Rust, Python, JavaScript, Java and PostgreSQL.
- WazoCore: Yoruba, Hausa and Igbo in a third fewer tokens (model on Hugging Face).
Open to collaboration on research and on building solutions that address important needs.