Fraud & Risk
Low-Latency, Web-scale Fraud Prevention with Samza and Friends
As eCommerce becomes at once more convenient for consumers and more complex for merchants, illegitimate account and order activity is becoming more and more challenging for merchants to pinpoint as vast amounts of data are typically involved. To help detect behavior indicative of fraud, modern fraud prevention systems need to process data with a high level of throughput and scale successfully as the business grows. The concept of assembly line processing is a great way to meet these challenges -- this presentation introduces streaming analytics through the Kafka ecosystem and Samza jobs, using sample output to help explain how data is processed in this type of system architecture.