Exams / AIF-C01 / ML lifecycle, MLOps and metrics / Question 3
AIF-C01 practice question: ML lifecycle, MLOps and metrics (3 of 4)
Fundamentals of AI and ML A free question from our original AIF-C01 practice exam. Try it, then open the answer to see why each option is right or wrong.
A small online shop has a product recommendation model that receives requests only a few times an hour, with long idle periods overnight. The shop wants to avoid paying for idle servers and does not want to manage instances or scaling. Which SageMaker AI inference option fits BEST?
Choose one answer.
- AA. Real-time inference on a persistent endpoint
- BB. Batch transform run once a night
- CC. Asynchronous inference with a request queue
- DD. Serverless inference
Show the answer and explanation
Answer: D
- A. Wrong: Real-time endpoints run on instances you choose for sustained traffic, so the shop would pay for them while idle.
- B. Wrong: Batch transform processes a whole dataset offline. The shop needs answers as requests arrive during the day.
- C. Wrong: Asynchronous inference suits large payloads and long processing times, which this workload does not have.
- D. Right: Correct. Serverless inference suits intermittent or unpredictable traffic, manages the infrastructure, and charges only for use.
Explanation
SageMaker AI offers four main ways to run inference. Real-time endpoints suit sustained, low-latency traffic. Serverless inference suits intermittent traffic with no idle cost. Asynchronous inference queues large or slow requests. Batch transform processes whole datasets offline with no persistent endpoint.
Study
Keep practising
Answer questions with a score at the end, or practise just this topic with the explanation after each one. No sign-up needed.