Skip to main content

Quick start

1. Prerequisites

Install these tools, sign in with az login, and create a GitHub token or Azure DevOps PAT. See the parameter checklist and prerequisites for details.

ToolVersionDirect download
PowerShell7.4 or newer (tested on 7.4.6). Windows PowerShell 5.1 does not work.PowerShell 7.4.6 releases
powershell-yaml module0.4.7 or newer0.4.7 on the PowerShell Gallery, or Install-PSResource powershell-yaml -Version 0.4.7
Terraform1.16.0 or newer (tested on 1.16.5)Terraform 1.16.5 downloads
Azure CLItested on 2.90.0; newer releases should workWindows installer, 2.90.0 release (macOS and Linux: install steps)
Bicepcurrent, only for iac: bicepaz bicep install, or the Bicep releases
gitcurrentgit downloads
GitHub CLI (gh)optional, 2.x (tested on 2.102.0)GitHub CLI releases

The Azure ML CLI extension (ml 2.45.1) is installed by the generated pipelines, so you do not install it yourself. Python 3.14 and the package versions are pinned in versions.lock.json and are used only by the pipelines.

2. Bootstrap

Import-Module ./src/MLOpsAccelerator/MLOpsAccelerator.psd1
Get-MLOpsScenario                                   # pick a scenario
New-MLOpsInputFile -Scenario managed-vnet-approved -Path inputs.yaml -Iac terraform -Cicd github
Test-MLOpsAcceleratorInput -InputFile inputs.yaml   # validate against the schema
Deploy-MLOpsAccelerator -InputFile inputs.yaml -OutputPath ./out            # generate only
Deploy-MLOpsAccelerator -InputFile inputs.yaml -OutputPath ./out -Deploy    # generate and bootstrap

3. Run

See Run for pushing the project, approvals, running each pipeline, local runs and cleanup.

Keeping current

Test-MLOpsAcceleratorVersion      # compare pins with latest releases
Update-MLOpsAcceleratorVersion

Terraform stays on azurerm 4.x because Azure Verified Modules require a version below 5.0; see versions.lock.json.

Phases of the generated project

  • Phase 0: platform prerequisites (outside the repository; private scenarios only).
  • Phase 1: bootstrap (terraform apply in bootstrap/).
  • Phase 2: infrastructure (infra pipeline, one environment at a time, with approvals).
  • Phase 3: workload (ml-train).
  • Phase 4: serve (ml-deploy and ml-promote).

The generated README has the full table with a stop-and-check step for each phase.