akhil324

Developed scalable fraud detection platform (Scala, Spark, GCP) processing multi-source transactional data, significantly reducing e-commerce losses. Integrated ML driven fraud detection models into Spark pipelines for proactive fraud mitigation (account takeovers, insider fraud, risky returns, linked accounts, etc). Optimized Spark & BigQuery pipelines, reducing query latency by 40% and cutting annual cloud costs by $50K+ through targeted optimizations. Migrated on-prem fraud analytics to GCP (Airflow, Spark), improving execution time by 3.5x and ensuring high availability. Integrated fraud analytics with React dashboards (Config-Driven Dashboards, Apache Freemarker, Spark) for real-time monitoring. Developed efficient data ingestion & transformation pipelines (clickstream, transaction logs, ML scores) supporting real-time and historical fraud analysis.

Engineered data pipelines ingesting diverse data (RDBMS via Sqoop/JDBC; NoSQL like CosmosDB/Cassandra; APIs via Python) in various formats (Parquet, Avro, JSON, XML). Utilized Hive and Spark (SQL/PySpark) for data transformation, optimizing Spark job performance by up to 350% through partitioning and advanced data modeling.