Azure AI-900 Model Deployment

A hands-on project demonstrating model deployment and inference using Azure AI Foundry, built as part of the Microsoft Azure AI Fundamentals (AI-900) certification journey.

📋 Overview

This notebook connects to a model deployed on Azure AI Foundry and sends chat completion requests using the OpenAI-compatible SDK. It demonstrates how to interact with cloud-hosted AI models through Azure's inference endpoint.

🛠️ Tech Stack

Component	Technology
Language	Python
Model Hosting	Azure AI Foundry
Model	Phi-4-mini-reasoning
SDK	OpenAI Python SDK (Azure-compatible endpoint)
Config Management	python-dotenv
Environment	Jupyter Notebook

⚙️ How It Works

1. The notebook connects to a model endpoint hosted on Azure AI Foundry.

2. Using the OpenAI-compatible client, it sends a chat completion request to the deployed model.

3. The model (Phi-4-mini-reasoning) processes the request in the cloud and returns a response.

4. API credentials are loaded securely from environment variables, keeping secrets out of the codebase.

🖥️ Run Locally

Clone the repository:

git clone https://github.com/osamag13/Azure-AI901-Model-Deployment.git

cd Azure-AI901-Model-Deployment

Install dependencies:

pip install -r requirements.txt

Add your Azure AI key:

Create a .env file in the root directory and add:

AZURE_AI_KEY=your_azure_key_here

Run the notebook:

jupyter notebook "Model\_Deploy-Osama Ghafoor.ipynb"

📚 What I Learned

• Deploying and calling models hosted on Azure AI Foundry

• Using the OpenAI-compatible SDK to interact with Azure-hosted endpoints

• Secure credential management using environment variables

• Practical, hands-on application of concepts covered in the AI-900 certification

📄 License

This project is open source and available for anyone to learn from and build upon.

---

Built by [Osama](https://github.com/osamag13) as part of Azure AI-900 certification preparation.