Skip to content

Azure AI Solution Design Guide

🔍 Problem

You need to design and develop a comprehensive AI solution using Microsoft Azure, which involves choosing the right services, tools, and architectural patterns to support capabilities such as generative AI, computer vision, speech, and natural language processing — while ensuring responsible AI practices.

✅ Solution with Azure

Use Azure AI services and the Azure AI Foundry platform to plan, develop, and manage AI-powered applications. Foundry supports generative AI development, agent-based workflows, and orchestrations like Prompt Flow, enabling robust, scalable solutions.

🧩 Required Components

  • Azure AI Services: Vision, Speech, Language, Translator, OpenAI, Document Intelligence, Content Safety, Face, Custom Vision
  • Azure AI Foundry (portal + SDK)
  • Azure AI Hub (optional) for advanced workflows
  • Development Tools:
  • VS Code (local or container image in browser)
  • Azure CLI / Bicep / ARM for provisioning
  • GitHub + GitHub Copilot
  • SDKs and APIs:
  • Azure AI Foundry SDK
  • Azure AI Services SDKs
  • Prompt Flow SDK
  • REST APIs

🏗️ Architecture / Development

Azure AI Foundry Structure

  • Foundry Projects:
  • Centralize development of chat apps and agent-based solutions
  • Connect to Azure AI services
  • Hub-based Projects:
  • Include managed compute, storage, key vault
  • Ideal for Prompt Flow, fine-tuning, multi-role teams (developers, data scientists)

Provisioning Options

  • Single-service resources: Ideal for small apps or evaluation (includes free tier)
  • Multi-service resource: Easier management across services like OpenAI, Speech, Translator, Vision, etc.
  • Important: Select the correct Azure AI Services resource type (🧠 icon), not legacy Cognitive Services

Developer Environments

  • VS Code container (pre-configured for Foundry)
  • Visual Studio / VS Code with GitHub integration
  • SDKs available for: C#, Python, Node, Java, TypeScript

✅ Best Practices / Considerations

Responsible AI Principles (Microsoft):

  • Fairness: Avoid bias, ensure representativity in training data
  • Reliability & Safety: Use thresholds, validate confidence scores, rigorous testing
  • Privacy & Security: Protect sensitive data during training/inference
  • Inclusiveness: Diverse team input during development
  • Transparency: Communicate system limitations, decision logic

Additional Considerations:

  • Regional Availability: Check availability per service/model
  • Cost Management: Use pricing calculator and Foundry's centralized cost control
  • Quota Awareness: Especially for VS Code compute containers and model usage
  • Tool Selection: Match IDE and SDKs to your language and workflow preference

❓ Sample Exam Questions

Q1: You are planning to build an AI application that extracts key information from scanned receipts. Which Azure service should you use? A1: Azure AI Document Intelligence

Q2: What is the benefit of using Azure AI Foundry instead of provisioning services individually? A2: Centralized project/resource management, integrated portal + SDKs, simplified deployment of multi-service solutions

Q3: You need to develop a generative AI agent that can act autonomously based on prompts. What Azure components should you consider? A3: Azure OpenAI (via Foundry), Azure AI Foundry Agent Service, optionally Prompt Flow SDK

Q4: A teammate wants to build a model fine-tuning workflow and securely store data and secrets. Which type of project is most suitable? A4: Hub-based project in Azure AI Foundry

Q5: What is a key difference between the Azure AI services icon and the Azure Cognitive Services icon when provisioning in the portal? A5: The Azure AI services icon provides access to the latest services (e.g., OpenAI, Content Understanding), while the older Cognitive Services icon does not.

Q6: Your development team prefers working in-browser and wants pre-configured SDKs. What is the recommended approach? A6: Use the VS Code container image via Azure AI Foundry portal