MLA-C01 Question 270
Single answerYou are building a machine learning application for processing customer support tickets. The application must classify tickets into categories in near real-time with low latency requirements. Additionally, the system should be cost-efficient as the traffic volume fluctuates throughout the day. Which type of endpoint is the most suitable for this use case?
- A
Serverless inference endpoint
- B
Real-time inference endpoint
- C
Asynchronous inference endpoint
- D
Batch transform
Show answer and explanation
Correct answer: A
Explanation
The application needs to classify tickets in near real-time while handling fluctuating traffic volumes cost-effectively. Serverless inference endpoints meet these requirements by providing low-latency responses and automatically scaling infrastructure based on demand, ensuring cost efficiency. Other options either lack the real-time capability or are not cost-effective for fluctuating traffic.
- A. Correct.
Serverless inference endpoints are well-suited for use cases with fluctuating traffic patterns and low latency requirements. They automatically scale the underlying infrastructure based on demand, making them cost-efficient and ideal for the described scenario.
- B. Incorrect.
Real-time inference endpoints are designed for low-latency scenarios but require a persistent infrastructure, which may not be as cost-efficient when traffic fluctuates significantly. This makes them less ideal for the given use case.
- C. Incorrect.
Asynchronous inference endpoints are used for scenarios with high latency tolerance, where results can be processed later. This does not align with the near real-time requirements of the use case.
- D. Incorrect.
Batch transform is designed for performing inference on large datasets in a non-real-time, offline manner. It is not suitable for low-latency, real-time requirements.