Bringing Data and AI Into Banking Products, Carefully

Data wrangling is the unglamorous foundation

Before anyone talks about AI, there's the question of whether your data is even trustworthy. In banking it usually isn't, not cleanly. Years of wrangling data taught me that most "AI problems" are actually data problems wearing a fashionable costume. The model is rarely the hard part; getting reliable, reconciled, well-modelled data is.

So when I move into a risk and compliance product, my first instinct isn't to ask what model to use. It's to ask what we actually know, how confident we are in it, and where the gaps hide. AI on bad data just produces confident mistakes faster.

Compliance changes the rules for AI

I now lead product in Financial Services Risk and Compliance, and that domain has opinions about black boxes. A recommendation engine for shopping can be inscrutable; a model that touches risk or regulatory decisions cannot. Explainability, auditability, and traceability aren't nice-to-haves, they're table stakes.

That's a healthy constraint. It forces you to use AI where it genuinely helps, surfacing anomalies, reducing manual operations, drafting and triaging, while keeping a human accountable for the call. The goal isn't to remove people from the loop. It's to remove the drudgery so people spend their judgement where judgement matters.

Automation should target the boring, repetitive work

Across my product roles I've consistently led automation of repetitive manual tasks, and the wins were real, meaningful reductions in manual operations and far more efficient use of people's time. AI extends that instinct rather than replacing it. The best place to point it is the same place: the dull, error-prone, repetitive work that quietly drains a team.

I'm an ML and AI enthusiast precisely because I've seen how much operational waste exists in financial services. You don't need a moonshot. You need to remove friction, one well-scoped task at a time.

Where I'm headed

Eighteen years of banking products plus a serious grounding in data and generative AI is a deliberate combination. I want to build financial products that are smarter without being reckless, using AI to cut the boring work and sharpen decisions, while keeping the precision, explainability, and trust that money has always demanded.