AI policy and ethics, informed by how AI systems actually break.

I have attacked retrieval-augmented language models in research and now work in AI regulatory research. My essays ask who is accountable when AI fails, what consent means when nobody reads the terms, and when human oversight exists only on paper.

Email me Read the essays

Recent writing

All writing

10 Questions on Humanity and AI I’m Still Struggling With

If AI saves us time, what are we doing with the time it saves?

16 Oct 2026

AI as Infrastructure: Open Questions on Reliance and Harm

We are trying to govern a technology while our relationship with it is still being negotiated.

11 Sep 2026

Formal Control, Real Dependence: Human Oversight and Automation Bias

A person can stay formally in control of a decision while losing the confidence to judge it.

4 Sep 2026

Who governs AI when everyone can build it, live session with Data Fellows

As building with AI gets easier, more people take consequential actions with it. Four questions to answer before you build or deploy anything with AI in it.

19 Sep 2026 · Essay and live session

Read the essay Watch the session

Solving overlooked problems.

Work With Me

Research interests

Evaluation capacity in countries that import AI

Most AI systems used in Nigeria are built elsewhere. I am working on what a country needs in order to test those systems itself, before and after deployment, instead of relying on the developer's own claims.

Under way.

Adversarial attacks on RAG systems

I built a hybrid retrieval pipeline (FAISS, TF-IDF, cross-encoder reranking, FLAN-T5), attacked it with synonym substitution and word deletion, and scored the damage with an Adversarial Risk Index I developed.

Completed as my undergraduate thesis.

Read the project

How Nigerians talk about AI

A study of public discussion of AI in Nigeria, using a data collection service I built for Bluesky that can be extended to news and video platforms.

Under way.

Product Strategy & Innovation

I build products for markets that are usually served last. Quotla handles business paperwork for owners without technical staff, and Testifi gives people control over how their personal stories are shared.

Learn More

AI Governance & Evaluation

I study how AI systems can be tested and held to account after they leave the lab, especially where they are bought rather than built. My work draws on security evaluation and on frameworks such as the NIST AI RMF and the EU AI Act.

See my research

Design Thinking

I design brand systems and visual explanations, so that technical work can be understood by the people who have to act on it.

Start This Engagement

Speaking

I have held leadership roles in organisations of different sizes, in agriculture, confectionery and technology. I speak about what those roles taught me, and about AI for audiences who have to make decisions about it.

Learn More