TechPrufer
Sign inGet started

Blogs›Redis›Real-time Fraud Detection

Real-time Fraud Detection

Real-time Fraud Detection

1
posts
2026

This post details the architecture and techniques for building a real-time fraud detection system, emphasizing latency constraints, feature stores, and the use of specific data structures for velocity calculations. It highlights how Redis can serve as the low-latency inference layer, supporting both exact and probabilistic counting methods for various fraud signals. The discussion covers scaling challenges, the importance of separating training and inference data stores to avoid training-serving skew, and the benefits of using sliding-window data structures like sorted sets, Bloom filters, HyperLogLog, and Count-Min Sketch for efficient feature computation.

2026

Real-Time Fraud Detection: Latency, Features & Scale

6/10/2026

This post introduces the concept of real-time fraud detection as a latency-bound problem, detailing the critical time constraints imposed by payment authorization flows. It explains the role of feature stores in providing necessary context for fraud models, contrasting the needs of training and inference data stores and addressing the training-serving skew problem. The post specifically highlights the use of Redis for the online inference layer and discusses the application of sliding-window data structures (sorted sets, Bloom filters, HyperLogLog, Count-Min Sketch) for calculating velocity features and other signals efficiently at scale.