Skip to content

📅 AI-102 Study Guide — Application Development with Azure OpenAI

✅ 1. Problem (Use Case)

You need to develop an application that:

  • Understands natural language
  • Generates personalized responses
  • Maintains conversation context (chat history)
  • Can generate promotional emails, categorize articles, and interact with provided content

🧬 2. Azure Solution

Use Azure OpenAI Service and GPT models (e.g., GPT-4o) through:

  • SDK (Python or C#) or REST API
  • Prompt engineering to control tone, format, and content
  • Grounding context for responses based on external sources

🔧 3. Required Components

Component Description
Azure OpenAI Resource To be created in Azure portal
GPT-4o Model Deploy from Cloud Shell
Azure.AI.OpenAI SDK C# (.NET) or Python (openai)
Keys/API Endpoint Available in Azure OpenAI resource
.env or appsettings.json file To configure endpoint, key, deployment
Prompt + System Message To control model response
Grounding context (optional) For responses based on provided textual context

🛠️ 4. Architecture/Development

A. Provisioning

  • Create Azure OpenAI resource (regions: East US, West US, etc.)
  • Deploy model via Bash in Cloud Shell with:
az cognitiveservices account deployment create \
  -g <resource_group> \
  -n <service_name> \
  --deployment-name gpt-4o \
  --model-name gpt-4o \
  --model-version 2024-05-13 \
  --model-format OpenAI \
  --sku-name "Standard" \
  --sku-capacity 5

B. SDK Configuration

Python:

from openai import AsyncAzureOpenAI
client = AsyncAzureOpenAI(
    azure_endpoint=azure_oai_endpoint,
    api_key=azure_oai_key,
    api_version="2024-02-15-preview"
)

C#:

AzureOpenAIClient azureClient = new (new Uri(oaiEndpoint), new ApiKeyCredential(oaiKey));
ChatClient chatClient = azureClient.GetChatClient(oaiDeploymentName);

C. Execution with Prompt Engineering

Use different types of prompts:

  • Generic: "You are an AI assistant"
  • Formatted: "You help write promotional emails"
  • Specific: instructions on tone, content, table formatting

D. Grounding Context + Chat History

  • Load grounding.txt
  • Initialize chat with grounding text
  • Add system and user messages to historical context
  • Maintain conversation history for consecutive questions

🔧 5. Prompt Engineering (Examples)

Prompt 1

  • System: You are an AI assistant
  • User: Write an intro for a new wildlife Rescue

Prompt 2

  • System: You are an AI assistant helping to write emails
  • User: Write a promotional email for a new wildlife rescue...

Prompt 3 (with specific tone)

  • System: You are an AI assistant that writes in a light, chit-chat style...

Prompt 4 (grounding)

  • Grounding: text from grounding.txt
  • System: You are an AI assistant who helps people find information...
  • User: What animal is the favorite of children at Contoso?

📆 6. Exercise and Testing

  • Files to modify: system.txt, grounding.txt
  • Run via dotnet run or python application.py
  • Observe how responses change as the prompt varies

❓ 7. Possible Exam Questions

  • How do you connect an application to Azure OpenAI?
  • How do you influence the tone and content of a GPT model's response?
  • How do you use grounding context to improve responses?
  • How do you maintain chat history in a conversation with a GPT model?

Study guide for AI-102 Microsoft Azure AI Engineer Associate certification