Skip to main content

Azure

Polaris AI offers an official Terraform module for deploying confidential AI infrastructure on Microsoft Azure. This module provisions confidential computing containers optimized for AI model inference.

Accessing the Module​

The module is available on the Terraform Registry:

Terraform Registry - Fr0ntierX/polaris-gpu/azure

module "polaris_azure_gpu_module" {
  source = "../"

  subscription_id = "YOUR-SUBSCRIPTION-ID"
  name            = "polaris-ai"
  location        = "eastus2"
  zone            = "2"

  admin_username        = "azureuser"
  admin_password_or_key = "MY_PASSWORD"
  authentication_type   = "password"

  virtual_network_name           = "my-precreated-vnet"
  virtual_network_resource_group = "my-network-rg"
  subnet_name                    = "my-subnet"
  
  custom_workload = {
    image_address = "example.azurecr.io/custom-workload:latest"
    port          = 11434
    registry = {
      login_server = "example.azurecr.io"
      username     = "registry_user"
      password     = "registry_password"
    }
  }

  polaris_proxy_enable_input_encryption = true
  polaris_proxy_enable_output_encryption = true
}

Available Features​

FeatureDescription
Confidential ComputingAzure Container Instances with confidential computing support
AI Model OptimizationsConfigurations for efficient AI model inference
Secure Data ProtectionOptional Azure Key Vault integration for sensitive data
CPU/GPU Split ArchitectureSecure preprocessing in TEE with GPU acceleration for inference

Learn More​

For detailed configuration options and usage examples, refer to: