Hannah A prediction service allows applications to send input data to a trained model and receive predictions. What components are needed to build a reliable prediction service, and how should teams manage scalability, latency, and monitoring?