Three infrastructure tiers
Available tiers
We offer three pre-built infrastructure tiers, each with consistent provisioning and access management to suit different stages of the machine learning lifecycle.
- Experimentation – A lightweight environment designed for early-stage development, enabling users to log experiments, track progress, and assess project feasibility without committing to full production infrastructure.
- Simple end-to-end – An environment with a complete workflow for teams ready to operationalize models, collaborate on results, and deploy to a secure and isolated production workspace.
- Full end-to-end – Includes separate development, test, and production workspaces. Recommended when changes must be validated before going live or when models support downstream applications.
Feature matrix for different tiers, three different versions in the columns
The primary differences between the tiers relate to the number of environments provided and the number of the deployment workflow.
What the tiers provide
Across all tiers, AI Platform's managed Azure ML provides:
- Automatic creation of Azure ML resources, storage accounts, and other dependencies
- Automatic setup of user groups and role assignments
- Controlled access management aligned to Equinor’s governance model
- A ready-to-use template repository for project setup and workflow execution
info
You can always upgrade to a higher infrastructure tier as your project requirements evolve. This flexibility ensures you can scale your environment and capabilities without disruption to your workflow.
Workflow examples
You can find workflow examples used in the tiers in the following repositories: