How azure-samples/azure-ai-search-with-content-understanding-python works
Project type and purpose
• This is a “full-stack sample” repository demonstrating how to ingest, analyze, and index multimodal content (documents, images, audio, video) using Azure AI Content Understanding and Azure Cognitive Search. It also provides:
• Jupyter notebooks (Python) to prototype document/video/layout extraction and RAG workflows
• A Next.js/React-based video-search web application
• Infrastructure-as-code (IaC) in Bicep to provision Azure resources
• CI/CD pipeline definition (azure.yaml)
• Analyzer templates (JSON) defining various Content Understanding pipelines
File structure
• Top-level folders:
• infra/ – Bicep modules and scripts that declare Azure Search, AI services, roles, and environment file writers
• notebooks/ – sample Jupyter notebooks illustrating different Content Understanding scenarios
• python/ – minimal Python client wrapper for Content Understanding API (requirements.txt)
• nodejs/video-search-app/ – a Next.js app that calls the Content Understanding API and queries Azure Search
• analyzer_templates/ – JSON configs parameterizing call recording, document, image-chart, and video pipelines
• data/ – sample media files used in notebooks
• azure.yaml – CI/CD workflow for building/deploying samples
• .devcontainer/ – dev-container configuration for GitHub Codespaces or VS Code
Read the README for additional insights
• Confirms use of Azure AI Content Understanding (2024-12-01-preview)
• Shows four primary notebook use cases (multimodal RAG, layout, visual document, video)
• Recommends Azure Developer CLI (azd) for spinning up resources or manual resource creation