Replacing Legacy Fraud Detection with an AWS ML Architecture

Software Data & Ai
Written By Josh Mackey

Overview

ImagineX partnered with a leading merchant payment platform to assess and modernize their payment fraud detection capabilities. After identifying critical gaps in their legacy Splunk-based solution, ImagineX designed a cloud-native AWS Lakehouse architecture powered by Amazon Fraud Detector and SageMaker to deliver ML-driven transaction monitoring.

Download Story
 
 

Problem

  • Legacy Splunk-based fraud detection had poor analytics architecture and limited machine learning capabilities

  • Missing data management and governance created significant gaps in fraud detection coverage

  • Alert fatigue reduced the Risk and Compliance team's ability to respond effectively to genuine threats

  • No centralized data infrastructure to consolidate transaction data from multiple sources

Solution

ImagineX assessed the existing fraud detection infrastructure, business rules, and monitoring workflows, identifying critical architectural gaps. ImagineX then designed a Lakehouse Architecture using AWS Redshift as the consolidated data warehouse, feeding Amazon Fraud Detector and SageMaker for ML-driven anomaly detection and case management.

Outcome

  • 1000x Scalability Improvement: the fraud detection migration enabled the platform to scale transaction monitoring to meet enterprise demand while optimizing cost..

  • Real-Time Sub-30ms Scoring: Successfully implemented an MLOps infrastructure capable of real-time ML-based fraud scoring that adheres to a strict transaction SLA.

  • Alert Fatigue Reduction: By replacing manual Splunk-based rules with ML-driven anomaly detection, Risk & Compliance team saw a significant reduction in false positives.

  • Centralized Data Governance: Established a single "source of truth" for payment transaction data within AWS Redshift, closing critical gaps in fraud coverage and providing the foundation for future AI-driven business intelligence.

Services Delivered

  • Data & AI Innovation

  • ML Fraud Detection Architecture

  • Data Lake & Warehouse Architecture

  • Data Engineering

  • Technology & System Architecture

Engagement Team

  • Engagement Lead

  • Solution Architect

  • Data Engineer

  • ML Engineer

  • Delivery Manager

Technologies Used

  • AWS

  • Amazon Sagemaker

  • Amazon Redshift

  • Amazon Fraud Detector

 
Josh Mackey
Previous

Achieving PCI DSS Compliance Through AWS Cloud-Native Security
Next

Global Payment Ecosystem Transformation for International Scale