PRODUCTION SCENARIO
A hospital network uses Amazon SageMaker AI to run a model that analyzes radiology scans of about 700 MB each. Each scan takes approximately 8 minutes to process. Clinicians need results within 1 hour of submitting a scan.
Which Amazon SageMaker AI inference option will meet these requirements?
Answering here is anonymous. Nothing is saved unless you sign in.
Show answer and explanation
Answer: Asynchronous inference
Asynchronous inference queues incoming requests and is built for payloads up to 1 GB, processing times up to one hour, and near real-time latency, and it scales the instance count to zero between scans. Real-time and serverless inference must both return inside the request timeout and accept much smaller payloads. Batch transform runs over an assembled dataset, so a scan waits for the next job rather than its own deadline.