Implementation and Integration II
Explore advanced deployment and integration methods for generative AI applications on AWS. Understand how to design scalable solutions using Amazon Bedrock, SageMaker endpoints, and managed services such as Step Functions to handle workflow orchestration, optimize GPU inference, and enable event-driven processing. This lesson equips you to efficiently integrate GenAI into existing enterprise systems with reliable and performant architectures.
We'll cover the following...
Question 31
A company is deploying a GenAI-powered internal knowledge assistant for employees. The assistant handles a high volume of simple requests, such as intent classification, query routing, and FAQ lookups, as well as a smaller number of complex requests that require deep reasoning over large internal policy documents and technical manuals. During business hours, the assistant must deliver consistent response times for complex reasoning queries, even when traffic spikes, and it must avoid managing custom model hosting infrastructure. The solution must integrate cleanly with existing application services and avoid unnecessary infrastructure management.
Which deployment strategies best meet these requirements? (Select any two options.)
A. Use Amazon Bedrock provisioned throughput for the large reasoning model.
B. Route all requests to a single large foundation model to simplify architecture.
C. Implement model cascading to ...