I build pricing and risk engines at BNP Paribas, and products of my own.

I’m the team’s only C# engineer, and the engines run on CPU and GPU. I’ve won the bank’s digitisation competition and a Take Ownership award.

Work

History, then marketing, then code. Each move was towards the work I liked most.

Building a whole product well: what to make, why it works, and getting it to production. How far I’ve got on each.

Products of my own, from a rental calculator to a risk simulator.

Four habits, and where I picked them up.

Start from a real problem.Threshold started because my friends had no clear idea what renting would cost by area. Flowtime started because Pomodoro kept cutting me off mid-flow.
Write it down first.Before I build, I write down the problem, who it’s for, and what’s out of scope. Threshold had 57 user stories before any code.
Ship it so it’s safe to undo.At BNP Paribas, taking a stale data dependency out of live pricing, I put the new path behind a feature flag that defaults to off, with caching and an end-to-end test nobody had asked for.
Measure whether it works.I learned this in marketing. At Optegra, split tests and analytics dashboards showed which changes brought in consultations and guided what the engineers built next; consultation conversion rose by 15%.
Threshold’s PRD16 April 2026

What my own time looks like this week.