1. What You’re Being Tested On
Using Azure AI Foundry, you should recognise:
How a model is deployed
How to interact with it (test it)
Where this fits in the AI solution lifecycle
👉 Think:
“Make the model usable, then try it out.”
🚀 2. Deploying a Model (High-Level Steps)
1️⃣ Select a Model
Choose from the model catalogue:
GPT / LLM → text generation
Vision → images
Speech → audio
2️⃣ Create a Deployment
You configure how the model will run:
Key ideas (exam level):
Give the deployment a name
Choose capacity/compute (abstractly)
Set configuration options
👉 Result:
The model becomes accessible via an endpoint (API)
3️⃣ Model is Now “Live”
Once deployed:
It can be called by apps
It can be tested in the portal
💬 3. Interacting with the Model in Foundry
After deployment, you use the playground / chat interface in
Azure AI Foundry
🧪 What You Can Do
✍️ Enter Prompts
System prompt → define behaviour
User prompt → ask questions
⚙️ Adjust Parameters
Temperature (creativity)
Max tokens (response length)
👀 View Responses
Evaluate accuracy
Refine prompts
🔁 Iterate
Modify prompts
Re-test outputs
Improve results
🔄 4. Typical Workflow (Exam-Friendly)
Select model
Deploy model
Open playground
Enter prompts
Adjust parameters
Review responses
👉 This loop is critical for prompt engineering
🎯 5. Example Scenario (Exam Style)
Scenario:
“A developer wants to test how a chatbot responds before integrating it into an application”
Correct understanding:
Deploy model in Azure AI Foundry
Use playground to:
Enter prompts
Adjust settings
Evaluate output
⚠️ 6. Common Exam Traps
❌ Thinking deployment = training
✔️ Deployment = making model available
❌ Skipping interaction step
✔️ Testing in playground is key
❌ Confusing endpoint vs UI
✔️ Endpoint = for apps
✔️ Playground = for humans
🧠 7. Simple Memory Model
“Deploy → Test → Refine”
Deploy → make model available
Test → interact in portal
Refine → improve prompts
✅ Summary
To deploy and interact with a model in Foundry:
Use Azure AI Foundry
Steps:
Select model
Create deployment
Use playground to interact
Enter prompts and adjust parameters
Evaluate and refine