Bring Your Own Model Within the GitHub Copilot App
Alexander Elguezabal walks through bringing your own AI model into the GitHub Copilot app using Microsoft Foundry and Azure.
Overview
This episode explains why model choice and token limits matter, then demonstrates how to:
- Create Microsoft Foundry resources in Azure using the Azure MCP Server
- Deploy a model and obtain the endpoint and API key
- Configure Microsoft Foundry as a model provider inside the GitHub Copilot app
- Select the deployed model in Copilot and validate it works
- Diagnose and fix a token rate limit issue
What the episode covers (timeline)
- 0:00 Welcome and bring your own model overview
- 0:17 Why model choice and token limits matter
- 0:45 Model providers and hosting options
- 1:55 Microsoft Foundry and Azure for Students
- 2:39 Azure portal basics, credits, and costs
- 3:30 Creating resources with the Azure MCP Server
- 4:12 Installing Azure MCP and signing in
- 5:40 Creating and verifying Microsoft Foundry resources
- 7:58 Endpoints, API keys, and deployments
- 8:55 Deploying a GPT-5 mini model with Copilot
- 10:49 Connecting Microsoft Foundry as a model provider
- 11:46 Adding and selecting the deployed model
- 12:56 Fixing a token rate limit and testing again
- 14:21 Recap
Key steps demonstrated
1) Create Foundry resources with the Azure MCP Server
- Use the Azure MCP Server to create the required Microsoft Foundry resources.
- Sign in and verify the resources in the Azure portal.
2) Deploy a model and capture connection details
- Deploy a model (shown with GPT-5 mini).
- Identify the deployment endpoint and generate/copy the API key needed for secure access.
3) Configure GitHub Copilot to use your model provider
- Add Microsoft Foundry as a model provider in the GitHub Copilot app.
- Connect using the deployment endpoint and API key.
- Select the deployed model and run a quick test to confirm it responds.
4) Troubleshoot token rate limits
- Reproduce a token rate limit failure.
- Apply the fix and re-test to confirm the model works again.
Links and resources
- Series site: https://aka.ms/student-learning-series-website
- Episode artifacts (prompts/code): https://aka.ms/student-learning-series
- GitHub Copilot app info: https://aka.ms/StudentAI-GitHubCopilotApp
- Azure CLI install guide: https://learn.microsoft.com/cli/azure/install-azure-cli
- Azure signup: https://aka.ms/StudentAI-Azure
- Azure Education signup: https://aka.ms/StudentAI-AzureEducation
- Azure portal getting started video: https://aka.ms/StudentAI-AzurePortal
Presenter
- Alexander Elguezabal (Software Engineer at Microsoft): https://www.linkedin.com/in/alex-elguezabal/