Modulai built a transaction risk prediction model that processes millions of payments annually across thousands of e-commerce sites. The project covered feature engineering, model development, decision engine specification, and a full deployment plan, enabling the client to proactively contain financial risk across their payment network.
Outcome
30%
Increase in fraud capture
~1.5M USD
Monthly losses prevented
Challenge
The client is a large Fintech company in the payments space, providing an alternative to credit card payments. They process many millions of transactions yearly for thousands of e-commerce sites and online service providers. Each transaction can expose the involved parties to various financial risks, making effective risk containment crucial to the business.
Solution
The team worked with structured, tabular data of various types, focusing on careful feature engineering with implementability in mind. A gradient boosting model architecture was chosen to maximize prediction accuracy. The preprocessing and modeling pipelines were built with Python open-source libraries, and the proposed production implementation is based on managed cloud services. The project also included a full specification of the decision engine infrastructure and deployment plan.
Tools
The preprocessing and modeling flows were built using Python open-source libraries with a gradient boosting model architecture selected for maximum prediction accuracy. The proposed production implementation is based on managed cloud services, with a fully specified decision engine infrastructure and deployment plan.
Value created
The model scores transaction risk across a network handling millions of payments a year for thousands of merchants, letting the client flag and contain financial risk as transactions happen rather than absorbing losses after the fact. Built with implementation in mind from the start, the work went beyond the model itself to specify the decision engine and a full deployment plan, giving the client a clear path from prototype to production across their payment network.
Learn more
