How azure-samples/azure-ai-document-processing-samples works
Project type and purpose
• This is a “sample repository” that demonstrates how to build document-processing workflows using Azure AI services (Azure AI Document Intelligence, Azure OpenAI, Azure AI Language, Azure Translator) and Infrastructure-as-Code (Bicep).
• The repository contains:
• A Dev Container for local/ Codespaces development
• A PowerShell script + Bicep templates under /infra to provision Azure resources
• Sample code and Jupyter notebooks (Python & .NET) under /samples showing classification, extraction, and redaction scenarios.
File structure
• Top-level directories:
• .devcontainer – development environment configuration (Docker, VS Code settings, tool installs)
• infra – Bicep templates and deployment scripts to stand up Azure AI Foundry, AI Services, Storage, Key Vault, Monitoring, Container Registry, etc.
• samples – language-specific folders with notebooks and helper modules for each scenario (classification, extraction, redaction)
• Configuration: .env.template, Setup-Environment.ps1, Bicep config files, GitHub CI files, CODEOWNERS, CONTRIBUTING.md.
Key architecture points from README
• Dev Container is used to unify tooling and dependencies.
• Setup-Environment.ps1 calls Azure CLI and Bicep to deploy:
• Azure AI Foundry Hub & Project (with serverless GPT, Phi endpoints)
• Azure AI Services (OpenAI, Document Intelligence, Language)
• Azure Storage (for artifacts)
• Azure Monitor (for logs)
• Azure Key Vault + Managed Identity (for secret management)
• Azure Container Registry (for any required containers)
• Sample code runs in that environment and calls AI service endpoints, using credentials stored in Key Vault via Managed Identity.
Detail
• Every IaC module folder under infra (ai_ml, storage, security, governance, containers)
• How key Bicep parameters (main.parameters.json) feed into the deployment
• Authentication flow (Managed Identity → Key Vault → AI Services)