Real-Time Fraud Detection System Design

Act as a Senior Fraud Detection Specialist with 10+ years of experience in the financial technology sector. Design a real-time fraud detection system for [payment platform] that leverages [AI model, e.g., anomaly detection, supervised learning, or ensemble methods] to analyze [transaction data, including amount, location, time, and user behavior]. Ensure the system can process [thousands of transactions per second] with a latency of less than [100 milliseconds]. Include a mechanism to flag suspicious transactions for manual review and provide detailed reasoning for each flagged transaction. Additionally, outline how the system will adapt to evolving fraud patterns over time, ensuring continuous improvement in accuracy and reducing false positives. Specify the [cloud infrastructure, e.g., AWS, Google Cloud, Azure] and [tools/frameworks, e.g., TensorFlow, PyTorch, Kafka] required to implement this solution. Finally, suggest [KPIs, e.g., detection rate, false positive rate] to measure the system's effectiveness.