Detecting Fraud Rings with GraphML: Real-Time Entity Scoring Using Neo4j and GNNs

Session Track: Data Intelligence

Session Time: 04:30 am - 05:00 am November 6

Session description

In this session, Akash Chandra, founder and CEO of InsightAI, will explore how graph machine learning (GraphML) can be applied to detect fraud rings and high-risk behavior within financial transaction networks. Using Neo4j as the core data platform, he will demonstrate how real-world financial entities—including users, accounts, and transactions—can be modelled as dynamic graphs and enriched with behavioural and structural features. You will learn how GNNs, such as GraphSAGE, can be used to assign real-time risk scores to nodes and uncover hidden fraud patterns that evade traditional rules-based systems. Attendees will gain hands-on knowledge of graph data modeling for fraud analytics, constructing ML-ready subgraphs, and running graph-based inference pipelines. The session will also cover lessons from production deployments across banks and fintechs in India and the Middle East. You will walk away with practical patterns, Cypher queries, and architecture techniques to integrate Neo4j and GNNs into real-time fraud detection systems.