To modernize the bank's mainframe systems, Edstem proposed a scalable, cloud-ready microservices design. Services were built with Java Spring Boot and communicated via message queues. Initially, the design used a Choreography pattern, but after client feedback, it was revised to an Orchestrator pattern to support synchronous workflow. Both versions were presented and well-received by the client's technical architects. The POC demonstrated a single transaction type from end to end. Transactions were read from input message queues as fixed-length messages, parsed into Java POJOs, enriched with additional data (mocked for POC scope), and transformed to Falcon's required format. The POC then sent transactions to Falcon scoring server via REST endpoints, throttled to 300 TPS. Failed transactions were stored in PostgreSQL for retry, and the Falcon response was processed to extract decision codes (Approve, Decline, Quarantine). Final output messages were written back to output queues. This solution proved the architecture’s feasibility and validated real-time fraud detection using microservices, message queues, and Falcon integration.

Self-Service Analytics

Empower every team member to access, analyze, and visualize data without technical expertise

Real-Time Updates

Transactions are processed in real time using Falcon scoring, with throughput throttling and retry handling to ensure reliability. This supports high-volume, synchronous transaction processing.

< 100ms latency

Enterprise Security

Bank-grade security with role-based access control and comprehensive audit trails

• SOC 2
• GDPR