Building a generative AI application on Kubernetes

In this section, we will build a generative AI application with Streamlit. A diagram representing the architecture for this application is shown in Figure 11.3. In this application, the user will be able to choose which foundational model they are going to talk to.

Figure 11.3 – Foundational models’ application architecture

Let’s start with the Python code for the application. The complete code is available under the Chapter 11/streamlit-claude/app folders on GitHub. We will walk through the code, block by block:

1. Create a folder named app and inside it, create a main.py code file. First, we import the necessary files and create a client to access Amazon Bedrock runtime APIs:

   import boto3
   from langchain_community.chat_models import BedrockChat
   from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
   bedrock = boto3.client(service_name='bedrock-runtime...