Scope adding an AI Agent for managing Azure workloads

Creating a job requires the following steps:

• Authenticate with Azure Identity
• Upload configuration and/or input data to Blob Storage
• Create a batch pool with mounted containers
• Define a Docker container and deploy to Azure Container Registry
• Define a set of tasks with specific dependencies and exit conditions
• Run the job

All these tasks make API calls to Azure Batch, Azure Blob or Azure Container service.

These can be automated with the help of agent that uses an LLM model to understand a goal (e.g. Create a job that runs the code from my_folder).

The agent can use an Agentic AI framework (e.g Azure Foundry, OpenAI LLM models) to automate this workflow.

Implementation Details

To be provided later