Design a Payment Fraud Detection Service
Company: PayPal
Role: Software Engineer
Category: System Design
Difficulty: medium
Interview Round: Onsite
Design a fraud detection service for a payment platform. The service should evaluate payment attempts in real time and return a decision such as allow, deny, challenge, or send to manual review. It should use information such as transaction amount, merchant, user account history, device fingerprint, IP address, geolocation, payment instrument, velocity signals, chargeback history, and known fraud patterns. The design should support both rule-based decisions and machine-learning model scores, low-latency online evaluation, model and rule updates, auditability, analyst feedback, monitoring, and high availability.
Quick Answer: This question evaluates a candidate's ability to architect a scalable, low-latency payment fraud detection service that integrates rule-based decisioning and machine-learning scoring while addressing real-time processing, data pipelines, model and rule management, auditability, monitoring, and high availability.