This repository leverages Azure OpenAI's native On Your Data feature and integrates it with Streamlit, a robust framework for developing web applications purely in Python. The goal is to create an MVP for RAG with the most streamlined architecture possible.
This template has now been successfully included in the azd-templates Gallery aka.ms/awesome-azd.
Thanks to the Microsoft azd team. ❤️
- Optimal security architecture implementation with Managed Identity and RBAC Configuration in On Your Data.
- Automatically triggering of critical Ingestion Jobs API within On Your Data using Bicep Deployment Scripts.
- Significantly reduces the data preprocessing engineering effort with managed chunking & embedding processes in On Your Data.
- The chat interface requires only a single Azure OpenAI On Your Data ChatCompletion API, eliminating the need for complex interactions with Azure AI Search.
- RAG interactive web built with Streamlit framework in just 50 lines of pure Python code.
-
This repository has been optimized for GitHub codespaces. Please use the following badge to open a web-based version of VS Code in your browser.
-
Login to your Azure account:
azd auth login --use-device-code -
Provision Azure resources and deploy the application code:
azd up- Enter the environment name, Azure Subscription, and the location of the Azure resources one by one as instructed.
- This process involves uploading the test document
./data/employee_handbook.pdfto an Azure Storage Account and triggering the Ingestion Jobs task within On Your Data.
-
Setting Up a Python development environment and running web application:
cd app . start.shIf you are in GitHub codespaces, you will see the automatically opened preview browser on the right-hand side.
-
Redeploy the updated program:
azd deployRepeat steps 4 and 5 and enjoy your development process!
Here are some resources to learn more about the technologies used in this repository: