A payments company handling pay-in, pay-out and bulk disbursements
Real-time fraud detection on a data lake
A fraud-detection data lake for PayYantra Innovations, scoring pay-in, pay-out and bulk disbursements as they happen.
Background
The situation
Fraud signals for each payment sat with many parties, and every decision had to be made while the transaction was open.
- Fraud signals sat across gateways, switches, merchants, beneficiaries and partner banks.
- The same party could appear under different identities in different sources.
- Each decision had to be made while the transaction was still open.
Solution
Solution delivered
Zucol built one layered data foundation, with entity resolution and families of fraud rules scored in real time.
- Pay-in, pay-out and bulk disbursementintegrated across modes.
- Gateway, switch, merchant, beneficiary and partner-bank datain one model.
- Raw, curated and consumption zoneswith low-latency lookup.
- Entity resolutionconnecting the same party across sources.
- Families of fraud rules,versioned, backtested and shadow-run.
Outcomes
Results
Transactions are scored inline, and every rule is backtested before it goes live.
Inline scoring while the transaction is open.
Every rule backtested and shadow-run before going live.
Solution
Related solution
More client success stories
Asandas & Sons (HyFun)
A digital warehouse management system for office and warehouse-floor teams.
Read the story →Landmark One
A brand identity for a Gurugram commercial development, built on a single market insight.
Read the story →Purnea University
An exam backlog cleared, and a university that now runs on one system.
Read the story →Discuss your requirements with our team
Share a few details about your institution and the problem you want solved. A senior member of our team will get back to you.


