Case Study: Fraud.net achieves rapid, scalable fraud detection and $1M/week in customer savings with Amazon Web Services

Fraud.net - Customer Case Study

Fraud.net is a crowdsourced fraud-prevention platform that aggregates real-time data from thousands of merchants to combat rapidly evolving online payment fraud. Facing hundreds of new fraud schemes and variations daily, the company needed to build and retrain many targeted machine-learning models quickly while avoiding the time and cost of building its own backend infrastructure and ensuring extreme scalability as its data and customer base grew.

Fraud.net moved its stack to AWS—using DynamoDB, Lambda, S3, Redshift and Amazon Machine Learning—to streamline model development and deployment. The AWS-based solution halved model training time, reduced complexity, enabled continuous detection of new fraud patterns, and improved accuracy; Fraud.net now maintains sub-200 ms response times, runs 20+ models (with plans for hundreds), and helps customers avoid about $1 million in fraud losses per week.

Fraud.net

Oliver Clark

CTO