AWS Certified Generative AI – Professional (AIP-C01) — Question 46
An international company is building an AI assistant that uses RAG. The company wants the AI assistant to have near real-time, low-latency performance. The AI assistant must provide service to several geographic areas. The company's customers will use proprietary data with the AI assistant. The proprietary data must not leave the company's immediate geographic area.
Which solution will meet these requirements?
Answer options
- A. Deploy an Amazon Bedrock model with a cross-Region model inference profile. Create Amazon S3 buckets in each AWS Region the company operates in. Store a knowledge base in each respective S3 bucket. In each Region, configure Amazon Kendra to interact with the respective knowledge base. In each Region, configure an AWS Lambda function that uses Kendra and Amazon Bedrock to process AI assistant prompts.
- B. Deploy an Amazon Bedrock model in each AWS Region the company operates in. Configure an Amazon Bedrock cross-Region model inference profile. Configure a vector database that uses Amazon Bedrock Knowledge Bases. Store the knowledge bases in Amazon S3 in each Region the company operates in.
- C. Use AWS Outposts to deploy an outpost in each AWS Region the company operates in. Create Amazon S3 buckets to store knowledge bases in each corresponding Region. Deploy Amazon RDS configured as a vector database to each outpost. Deploy an Amazon Bedrock model with a cross-Region inference profile in a central Region.
- D. Configure a knowledge base stored in the Amazon S3 Express One Zone storage class in each AWS Local Zone the company operates in. Use Amazon RDS to deploy a vector database in each Local Zone the company operates in. Deploy a large language model (LLM) to Amazon EC2 instances in each Local Zone. Configure the AI assistant to route prompts to the model in the respective Local Zone.
Correct answer: B
Explanation
Option B is the correct choice because it allows the Amazon Bedrock model to be deployed in each AWS Region, ensuring low-latency performance while keeping proprietary data within the respective regions. Option A involves more complexity with AWS Lambda and Kendra for each region, which may not be necessary. Option C's use of AWS Outposts introduces additional infrastructure overhead, while Option D uses Local Zones, which may not meet the real-time requirement as effectively as the solution in Option B.