America First Credit Union

Case study

America First Credit Union

A shared data foundation for fraud monitoring, member recommendations, and regulatory reporting.

A banking staff specialist reviews a folder at a desk in a community branch.
Editorial illustration.

Fragmented banking data

America First Credit Union’s data was spread across mainframes, relational databases, and local reporting tools. That fragmentation constrained analytics and fraud monitoring, complicated regulatory audits, and limited personalized member outreach.

Fraud monitoring and member offers

The Emperitas team implemented an Azure data lakehouse using Databricks and Delta Lake, with role-based access controls and data masking. They established an enterprise data governance framework and built a near-real-time fraud pipeline that used streaming transaction features and gradient-boosted decision trees to score unusual activity.

They also developed a next-best-action model to recommend products, including auto loans and mortgage refinancing, within the online banking application.

Reported results

The project reported more than $4.5 million in fraudulent transactions prevented during the first full year of the automated fraud pipeline. Auto-loan campaigns using the recommendation engine reported a 3.2× increase in offer conversion rates.

Subsequent regulatory data compliance audits had no material findings, and preparation time for compliance reporting fell by 75%.