High-Throughput Transaction Security & Fraud Alerting

Operated and optimized a mission-critical fraud detection system processing 100k+ transactions per second (TPS). The system ingests payment card logs via a Java batch service, evaluating them against configurable risk rules.

Suspicious activities trigger enriched, parameterized SQL queries to validate threats with low latency before dispatching real-time SMS alerts to customers.

• Managed performance at extreme scale (100k+ TPS), utilizing Datadog to trace latency bottlenecks across the ingestion pipeline
• Optimized SQL query performance to prevent database locking during high-volume fraud spikes
• Reduced false positives by implementing complex rule-based logic validation steps