Let’s talk.

I work on the part of an AI deployment that happens around the model: fitting it into how a team already works, and deciding when the work is ready to rely on.

I fit forward-deployed and solutions roles, AI enablement and adoption, AI operations, and AI training and evaluation.

What I bring

• Invisible Three years at Invisible Technologies, the last of them as Training & Process Lead, running onboarding and quality for the teams that produce AI training data.
• The rubric and review workflow that brought an intake of more than a hundred people to production readiness in two weeks.
• Lagbe Naki The order system a sourcing business runs on, which I designed and built.
• Thinkaloud Thinkaloud, an open-source tool that makes the AI someone already uses show its reasoning.

My work in AI data production turned on getting trainees to show their reasoning clearly enough for a model to learn from it, and my B.Sc. thesis examined what shaped a malware classifier’s predictions. Both left me with questions I want to test properly in graduate research.

What I bring to a lab

• Checkable Reasoning Ten habits for checkable reasoning, written from three years of coaching them in AI data production.
• PDFGuardian A thesis on what a malware classifier’s verdicts rest on, re-examined in 2026 with the saved models and data.
• Invisible The rubric and review workflow behind an automated grader, and the human checks that followed its verdicts.

I start by learning how the work happens today, then define what a system needs to do before anyone picks a tool. The answer might be AI or a simpler system. If someone has already proposed a solution, I can judge whether it fits the problem.

Problems I’ve solved with technology

• Lagbe Naki Turned a sourcing business’s order flow into an app, from enquiry to confirmed profit.
• Rendered Audit Bridge Traced why an AI audit workflow kept overflowing its token limits, down to the page data its agents passed between them.
• DigiDesk Led the technical architecture of a design for tracking business applications across government offices.

I ran weekly sessions on showing your reasoning to AI and checking your own work before handing it over. A workshop brings the same practice to your group.

Sessions I can run

• Writing an AI policy for a course or department.
• Checking what AI gives you: the ten habits, practised on your own material.
• Building with AI safely: decisions you can defend, and a security check before launch.

One session or a series, for a small team or a full cohort. Plan yours on Thinkaloud