An ML model will score customers in an online application to determine which product to show. The engineer needs to minimize response latency. How should the model be deployed in SageMaker to meet low-latency requirements?

A real-time inference endpoint is the correct choice because it provides immediate predictions for individual requests, which is essential for low-latency requirements in online applications. Batch transform is for processing large datasets offline, not for real-time scoring. Serverless inference endpoints offer automatic scaling and cost-effectiveness but can introduce cold start latencies that might not meet strict low-latency demands. Asynchronous inference endpoints are designed for large payloads or long-running inferences where immediate responses aren't critical, making them unsuitable for this scenario.