GenAI Inner Loop Workshop - Upskill your search with AI

In this repository, you can find code samples for the Azure AI Search, Prompt Flow and building RAG patterns for search. Repository also contains self-paced, hands-on labs for the workshop. The workshop is designed to help you understand how to use AI to improve search relevance while Prompt Flow is a tool that helps you build high-quality LLM apps - from prototyping, testing to production deployment and monitoring. The workshop also covers inner loop details regarding environment setup, minimum viable infrastructure, and best practices for building AI-powered search experiences.

Prerequisites

Following steps will guide you through the setup of the environment and Azure resources for the code samples and the workshop.

• Azure CLI: Infrastructure is setup using Azure CLI. See detailed instructions on how to setup here.
• Azure Developer CLI: Azure Developer CLI is used to deploy the infrastructure. See detailed instructions on how to setup here.
• Python: Code samples are written in Python. Install Python from here. Examples require Python 3.9+.. You can also open the repo using DevContainer to setup the environment.

Setup Azure resources

You need to be already logged in to Azure CLI and Azure Developer CLI in your envrionment. If not, you can login using the following commands:

az login
azd login

1.Portal Access:

Set AZURE_PRINCIPAL_ID environment variable to be used setting up the portal resource access:

export AZURE_PRINCIPAL_ID=$(az ad signed-in-user show --query id -o tsv)

2. Deploy services with Azure Developer CLI:

This deploys infrastructure defined in infra folder using Azure Developer CLI. This includes Azure AI Search, Azure Storage, Azure Key Vault and related services required for the workshop and code samples.

azd up

3. Set environment variables for local development into .env file:

This command sets up environment variables required for local development into .env file. Notebooks and code samples can read from this file for API Keys and other secrets. .env is ignored in .gitignore file.

azd env get-values > .env

4. Setup python virtual environment:

This command sets up a python virtual environment and installs all the required packages from requirements.txt.

This command also sets up .gitattribute and nbstripout for Jupyter notebooks. This avoids the notebook outputs to be committed to the repository. nbstripout is a tool that can remove the output from Jupyter notebooks. When you run a Jupyter notebook, the output of each cell is saved in the notebook file. This can be problematic when using version control, as the output can change each time the notebook is run, even if the code hasn't changed. This can lead to large diffs and merge conflicts.

This installs Prompt Flow to build high-quality LLM apps - from prototyping, testing to production deployment and monitoring.

make setup-local-env

To use the newly created virtual environment, you can activate it using the following command:

source .venv/bin/activate