Talk at DevStandup

Episode #27 of DevStandup features Ricardo Mello, Senior Developer Advocate at MongoDB, breaking down how to build a real-time fraud detection system in Java—without training your own ML model.

This session focuses on what most teams get wrong: overengineering fraud systems when simple, fast pipelines and good heuristics already solve 80% of the problem. In 'Real-Time Fraud Detection in Java with Kafka, Streams & Vector Similarity', Ricardo shows how to ship a system that actually works under pressure.

This episode is for developers who want working systems, not theoretical architectures that never make it to production.

The full project code — including the Kafka Streams pipeline and MongoDB Vector Search integration — is available on GitHub.

You can watch the full episode on YouTube.