Overview
Dfinder FDS is a cloud-native fraud detection system re-architected for AWS by Interezen and BSG. Proven in demanding legacy environments, it now delivers real-time transaction analysis at cloud scale -- instantly flagging anomalies such as voice phishing, mule-account abuse, and fraudulent payments with sub-500ms response times.
The detection engine leverages Elasticsearch, Redis, and MariaDB for high-speed in-memory analysis the moment a transaction event arrives. A dynamic rule-based engine lets administrators compose detection rules from IP reputation, device fingerprints, and user transaction patterns, applying changes live through the management UI with no system restart required.
Dfinder FDS runs as containerized agents across an Amazon EC2 in private subnets. A Multi-AZ cluster fronted by an Application Load Balancer ensures zero-downtime availability, while Auto Scaling Groups expand capacity automatically as traffic surges.
Enterprise security compliance is built in: least-privilege IAM role control, EBS volumes encrypted with AWS KMS, and IMDSv2 enforcement with IMDSv1 fully disabled. Streaming logs pipeline into Amazon S3 provides forensic analytics, and Amazon EventBridge enables downstream eventing.
A dedicated browser script collects device signals from the customer's PC/web environment, and a data-transfer component on the customer's application server forwards transactions to the FDS analysis server securely over REST APIs.
Built for industries where money moves -- banking and finance, e-commerce and platforms, and insurance -- Dfinder FDS monitors every transaction in real time. It preemptively blocks voice phishing and mule-account abuse in financial institutions, stops fraudulent orders before fulfillment in commerce, and surfaces organized fraud rings in insurance claim transactions.
Interezen Co., Ltd. provides the FDS detection engine, rule management, device fingerprinting, transaction analysis, and detection policy consulting. BSG, an AWS Advanced Partner since 2015, delivers EC2 and IaC-based deployment automation, AWS security compliance, Multi-AZ and Auto Scaling operations design, and integrated support.
Highlights
- Sub-500ms Real-Time Detection: In-memory analysis Redis, and MariaDB processes each transaction event the instant it arrives, guaranteeing fraud detection responses within 500 milliseconds.
- Dynamic Rule Engine with Live Deployment: Compose detection rules from IP reputation, device fingerprints, and transaction patterns -- apply changes instantly through the admin UI with no system restart required.