Gopalakrishnan Venkatasubbu

Streaming Intelligence: Real-Time Fraud Detection in Retail Payments Using Machine Learning

Abstract:

The rapid growth of digital retail payments has significantly increased exposure to sophisticated and evolving fraud patterns, requiring real-time, adaptive detection systems beyond traditional rule-based approaches. This talk presents a cloud-native, event-driven framework for real-time fraud detection that integrates machine learning with streaming data architectures.

The proposed system leverages technologies such as Apache Kafka for high-throughput data ingestion, real-time feature engineering, and low-latency model inference using ensemble learning methods including Gradient Boosting and Random Forest. Transactions are continuously evaluated within milliseconds, enabling dynamic risk scoring and immediate decisioning in production environments.

Experimental results demonstrate high detection performance, with ensemble models achieving over 98% accuracy and strong precision-recall balance, effectively identifying complex, non-linear fraud behaviors while minimizing false positives. The session also discusses practical challenges such as concept drift, model retraining, scalability, and explainability in enterprise systems.

Based on peer-reviewed research presented at international conferences through both oral and recorded formats, this work bridges academic innovation and real-world deployment, offering actionable insights for building scalable, intelligent fraud detection systems in modern payment ecosystems.

Profile:

Gopalakrishnan Venkatasubbu is a Cloud-Native Systems Architect and applied AI researcher with over 20 years of experience at pne of the largest home improvement retailers in the United States, The Homedepot Inc,designing real-time distributed systems and machine learning - driven fraud detection platforms for large-scale retail payment environments with . His work focuses on building event-driven, low-latency architectures that enable intelligent, real-time decisioning in production systems.

He is the author of peer-reviewed research on real-time fraud detection using streaming intelligence and machine learning, presented at international conferences through both oral and recorded formats. He is regularly invited to serve as a reviewer and Technical Program Committee member for IEEE and international conferences, and as a judge for global industry awards and leading U.S. universities, reflecting recognition of his expertise in evaluating technical innovation.

His contributions lie at the intersection of applied AI, scalable systems, and financial technology, with a focus on translating research into enterprise-grade solutions.