AWS SimuLearn is an online learning experience that pairs generative AI-powered
simulations with hands-on practice to help individuals learn how to translate
business problems into technical solutions through the simulation of dialog between
a customer and a technology professional.
AWS SimuLearn:
Create Trustworthy LLMs with a Vector Database
In this AWS SimuLearn assignment, you will review a real-world scenario helping a fictional customer design a solution on AWS. After the design is complete, you will build the proposed solution in a guided lab within a live AWS Console environment. You will gain hands-on experience working with AWS services, using the same tools technology professionals use to construct AWS solutions.
How does it work?
AWS SimuLearn is powered by generative AI that enables you to have life-like conversations with AI customers. Your responses to the AI are evaluated to help develop your soft skills like communication. and problem solving. As you provide the correct responses, a quiz agent with test you and you can also get help from Dr. Newton, the helper agent when you get stuck. Once you provide all of the correct responses for the solution you will move to building it and validating in a live AWS Console environment.
Simulated Business Scenario
A wine retailer has a large wine selection on their ecommerce website. When customers shop the website, the retailer wants to provide wine recommendations based on customer descriptions of their wine preferences. Generative AI models could be used to generate recommendations on the spot; however, these models sometimes hallucinate details that seem plausible but are incorrect. This results in annoyed customers and lost sales opportunities. The retailer wants to maintain the efficiency of AI generation while reducing hallucinated outputs, so they can use these tools without jeopardizing customer satisfaction and brand reputation.
Learning objectives
- Describe how a vector database is used to create a knowledge base.
- Determine how to improve the trustworthiness of an LLM by incorporating RAG.
- Demonstrate how to configure a vector database by using Amazon RDS for Aurora PostgreSQL.
- Describe how metadata filtering can be used to retrieve relevant results from a knowledge base.
- Investigate issues related to data ingestion and metadata formatting.
AWS SimuLearn: Create Trustworthy LLMs with a Vector Database
Services Used
AWS Secrets Manager, Amazon Bedrock, Amazon Relational Database Service (RDS), Amazon S3
