A company needs to use Retrieval Augmented Generation (RAG) to supplement an open source large language model (LLM) that runs on Amazon Bedrock. The company's data for RAG is a set of documents in an

Question

Options

• ACreate a pipeline in Amazon SageMaker Pipelines to generate a new model. Call the new model
• BConvert the data into vectors. Store the data in an Amazon Neptune database. Connect the
• CFine-tune an existing LLM by using an AutoML job in Amazon SageMaker. Configure the S3
• DCreate a knowledge base for Amazon Bedrock. Configure a data source that references the S3

• A
  13% (3)
• B
  4% (1)
• C
  4% (1)
• D
  78% (18)

Explanation

Option D is correct because Amazon Bedrock Knowledge Bases is a fully managed, native RAG solution that natively supports S3 as a data source, handles document ingestion (including .csv and .docx files), automatically chunks and embeds documents into a vector store, and integrates directly with Bedrock-hosted LLMs - all with zero infrastructure to manage.

Why the distractors fail:

• A is wrong because SageMaker Pipelines is for training/building new models, not for RAG retrieval augmentation, and it introduces significant operational complexity.
• B is wrong because manually converting documents to vectors and managing Amazon Neptune (a graph database) requires substantial custom engineering - Neptune is not purpose-built for vector similarity search in RAG workflows.
• C is wrong because fine-tuning bakes knowledge into model weights at training time; it is the opposite of RAG, which retrieves external knowledge at inference time. Fine-tuning also requires retraining whenever data changes.

Memory tip: Think "least operational overhead" = "most managed." Bedrock Knowledge Bases is AWS's purpose-built, fully managed RAG service - if the question mentions RAG + S3 + Bedrock, Knowledge Bases is almost always the answer. Fine-tuning ≠ RAG; RAG retrieves, fine-tuning trains.

No community discussion yet for this question.