Implementation and Integration II
Explore practical solutions for deploying scalable GenAI applications on AWS. Learn to handle real-world scenarios involving model deployment, workflow orchestration with Step Functions, GPU container design, and event-driven architectures. This lesson helps you understand how to optimize AI-powered assistants and systems for performance and operational efficiency.
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 ...