About Okunola

I don't think of a research contribution as finished when a paper is accepted. My PhD work on adaptive credit-card fraud detection only became interesting to me once it had to survive contact with real transaction data, in a system where the population trying to defeat it was adapting in real time. That's the standard I now hold research to: does it still work when the assumptions it was built on start to shift.

Concept drift is the thread that runs through almost everything I work on — fraud, biometric verification, even the Lassa fever prediction research funded through a TETFund NRF grant — because in each case, the phenomenon being modelled doesn't hold still. I'm drawn to problems where a model's first evaluation number is the least interesting thing about it, and where the harder, more useful question is what happens to that number six months into deployment.

That's also why responsible AI isn't a separate research interest for me so much as a constraint on all the others: a fraud system that can't explain a weight change to a compliance reviewer, or a biometric framework that degrades unevenly across populations, hasn't actually solved the problem it was built for.

Leadership, for me, has mostly meant showing up for structures other people depend on: coordinating curriculum quality and staff development as Head of Team (Computing) at QAHE, serving as a School Governor at Ruskin Infant School, and founding Techaxle.ng, an online community supporting data scientists, analysts, and researchers in Nigeria.

None of that is separate from the mentoring work — a community, a governance role, and a research-supervision relationship are all, in different ways, about making a system work better for the people inside it than it would if left alone.