Sravani Chowdary Kasaraneni

At JPMorgan Chase, I build and productionize real-time fraud detection models for payment authorization systems, improving detection performance by 15% in high-risk transaction segments while meeting sub-100ms latency requirements.

I've reduced false positives by 20% in high-volume card transactions through threshold tuning, feature engineering, and decision-policy calibration. I integrate model inference into Kafka and Spark streaming pipelines that score millions of transactions daily, and I own the lifecycle from feature engineering through deployment, monitoring, and retraining.

Previously at Neon IT Systems, I developed machine learning models for personalization and ranking, along with scalable batch and near-real-time inference pipelines. I built Python backend services and monitoring workflows for reliable model serving, data validation, drift detection, and performance tracking.

My work centers on fraud detection, large-scale feature systems, low-latency inference, and production ML reliability using PySpark, Kafka, Apache Spark, Docker, AWS, and CI/CD automation.