Project

Real-Time Healthcare Fraud Detection

Streaming claims pipeline with graph-based fraud detection and automated alerting.

← All work

What I built

  • —Kafka + PySpark streaming on Databricks over 50M+ claims; cleansing improved data quality by 45%.
  • —Graph-based fraud model at 92% accuracy, retrained via Airflow on SageMaker.
  • —Monitoring dashboard with Lambda alerts cut investigation cycle time by 60%.