Major project Banking & Fintech Case Studio

Real-time payment fraud scoring service

Inspired by publicly reported work at Stripe

Score each payment for fraud in milliseconds with rules plus a model, and give analysts a review queue.

The brief

Stripe has described building Radar: features computed across many payments, models retrained often, rules that merchants write on top, and manual review for the doubtful cases. Build your own fraud scoring service: engineer features across a customer's history, train a model, serve a score in real time, let an analyst add rules, and send the doubtful band to a review queue. Good looks like: AUPRC on a time-based split, response time under a stated target, rules and model working together, a review queue, and a monitoring view.

Read the case first

What Stripe has said in public about this kind of work, on its own pages:

Make it your own

Score a different kind of risk: fake sign-ups on a job portal, refund abuse at an online store, or coupon misuse. Write five rules an analyst at that business would want and show how they work with the model.

AITeky has no link with Stripe, which has not reviewed or endorsed this brief. The brief only draws on what the company has published, uses open data rather than its data, and is a learning exercise: don't present your project as Stripe's work or system.

What you will build

Engineer velocity and history features without leakage
Train and calibrate a model on a time-based split
Serve a score through an API within a latency target
Add merchant rules on top of the score
Send the doubtful band to a review queue with reasons
Monitor fraud rate, false declines and drift

Free, open data to build it with

It takes a minute with your email. AiTeky Pass (₹999/month or ₹9,999/year) is free for your first 30 days.