Professional Experience

Architected Payment fraud detection data pipeline on Azure Databricks, processing 2M+ monthly transactions using Apache Spark, achieving sub-second P99 latency through DataFrame API optimization and adaptive query execution.
 

Implemented fraud pattern detection using iterative graph processing on 400K+ transaction edges with Spark MLlib and GraphX, employing PageRank algorithm to identify fraud rings, resulting in 15% reduction in fraudulent transactions saving $4M annually.