Rule Backtesting System

Grab has evolved its fraud detection capabilities by developing Griffin, an in-house anti-fraud risk rule engine. This system allows data scientists and analysts to create, manage, and deploy complex rules through a web portal, moving away from manual rule implementation in code. Griffin handles data orchestration for predictions and executes rules efficiently, making billions of predictions daily with low latency. It supports dynamic datasets, complex logic with Python-based rule execution, and a cloud-agnostic Spark-based data pipeline for backtesting. The system also includes a web portal for managing jobs and downloading results, and integrates with Kafka for historical event storage.