Rohit036 / amazon-bedrock-kendra-rag-ibm

Amazon Bedrock RAG demo application using Kendra as knowledge base

RAG demo application using Streamlit, Amazon Bedrock, LangChain and Amazon Kendra

This project offers a sample front-end application built with Streamlit to showcase RAG using Amazon Bedrock, LangChain and Amazon Kendra. You can deploy this CDK project in your AWS account. Alternatively, you can run the Streamlit application on your local machine. See sections below for instructions.

Architecture

Prequisites

Make sure you have access to the Bedrock models before using this application. You can request access to the Bedrock models through the console:

Enable AWS IAM permissions for Bedrock

The AWS identity you assume from your notebook environment (which is the Studio/notebook Execution Role from SageMaker, or could be a role or IAM User for self-managed notebooks), must have sufficient AWS IAM permissions to call the Amazon Bedrock service.

To grant Bedrock access to your identity, you can:

Open the AWS IAM Console Find your Role (if using SageMaker or otherwise assuming an IAM Role), or else User Select Add Permissions > Create Inline Policy to attach new inline permissions, open the JSON editor and paste in the below example policy:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "BedrockFullAccess",
            "Effect": "Allow",
            "Action": ["bedrock:*"],
            "Resource": "*"
        }
    ]
}

If you are running the application from your local machine, make sure you have the following tools installed:

• AWS CLI
• Node.js
• IDE for your programming language
• AWS CDK Toolkit (v2.0 or later)
• Git, JQ, etc
• Docker

Deploy application in your AWS account

We recommend you use an AWS Cloud9 development enviroment for this project. See this Cloud9 setup guide.

Clone the repository and enter the project directory:

git clone https://github.com/Hantzley/amazon-bedrock-kendra-rag-demo.git
cd amazon-bedrock-kendra-rag-demo

Create a virtualenv on MacOS or Linux:

python3 -m venv .venv

After the init process completes and the virtualenv is created, you can use the following step to activate your virtualenv.

source .venv/bin/activate

Once the virtualenv is activated, you can install the required dependencies.

pip install -r requirements.txt

If your account is not yet boostrapped for CDK, execute the following command:

cdk bootstrap

At this point you can now list the stacks in the project.

cdk ls

You should see the following output:

GenAiRagVpcStack
GenAiRagKendraStack
GenAiRagWebStack

Deploy the application as a container on Elastic Container Services in our AWS account:

cdk deploy GenAiRagKendraStack GenAiRagWebStack

The GenAiRagKendraStack and GenAiRagWebStack depends on GenAiRagVpcStack. CDK will resolve that dependency and automatically deploy GenAiRagVpcStack first.

Copy the WebApplicationServiceURL from the output and paste it on your browser.

Run application on your local machine

Alternatively, you can execute the Streamlit application on your local machine:

cd web-app
streamlit run Home.py --server.maxUploadSize 50

The application should open in your browser.

Enjoy!

Clean up

To avoid unnecessary cost, you can destroy the resources used in the project:

cdk deploy --all

Screenshots

RAG Application

Documents upload

Documents list