Terraform AWS ARC Sagemaker Module Usage Guide¶

Introduction¶

Purpose of the Document¶

This document provides guidelines and instructions for users looking to implement the Terraform AWS SageMaker AI.

Module Overview¶

A comprehensive Terraform module for creating and managing AWS SageMaker Studio resources — including domains, user profiles, models, endpoints, endpoint configurations, and ML pipelines. This module follows AWS and Terraform best practices, supports advanced customization, integrates seamlessly with IAM and VPC settings, and is designed to enable scalable, secure, and production-ready ML workflows.

Prerequisites¶

Before using this module, ensure you have the following:

AWS credentials configured.
Terraform installed.
A working knowledge of AWS VPC, All SageMaker services (Studio, Pipelines, Canvas, etc.), and Terraform concepts.

Getting Started¶

Module Source¶

To use the module in your Terraform configuration, include the following source block:

module "arc-sagemaker" {
  source                 = "sourcefuse/arc-sagemaker/aws"
  version                = "0.0.1"
  name                   = "terraform-arc"
  create_endpoint_config = true
  create_model           = true


  primary_container = {
    image          = "683313688378.dkr.ecr.us-east-1.amazonaws.com/sagemaker-scikit-learn:1.0-1-cpu-py3"
    model_data_url = "s3://your-sagemaker-model-bucket-21-05-25/model/model.tar.gz"
    environment    = {}
  }

  production_variants = [
    {
      variant_name           = "AllTraffic"
      initial_instance_count = 1
      instance_type          = "ml.m5.large"
      initial_variant_weight = 1.0
    }
  ]
  tags = module.tags.tags
}

Refer to the Terraform Registry for the latest version.

Integration with Existing Terraform Configurations¶

Refer to the Terraform Registry for the latest version.

Integration with Existing Terraform Configurations¶

Integrate the module with your existing Terraform mono repo configuration, follow the steps below:

Create a new folder in terraform/sagemaker named .
Create the required files, see the examples to base off of.
Configure with your backend:
- Create the environment backend configuration file: config..hcl
- region: Where the backend resides
- key: /terraform.tfstate
- bucket: Bucket name where the terraform state will reside
- dynamodb_table: Lock table so there are not duplicate tfplans in the mix
- encrypt: Encrypt all traffic to and from the backend

Required AWS Permissions¶

Ensure that the AWS credentials used to execute Terraform have the necessary permissions to create, list and modify:

All SageMaker services (Studio, Pipelines, Canvas, etc.)
S3 data access and model artifacts
CloudWatch logging and monitoring
VPC and networking configuration

Module Configuration¶

Input Variables¶

For a list of input variables, see the README Inputs section.

Output Values¶

For a list of outputs, see the README Outputs section.

Module Usage¶

Basic Usage¶

For basic usage, see the examples folder.

This example will create:

SageMaker Studio Domain: Complete domain setup with VPC integration
User Profiles: Multiple user profiles with customizable settings
ML Pipelines: SageMaker Pipeline support with S3 integration
model: Flexible SageMaker Model creation with configurable containers
Endpoint Configurations: Fine-grained control of production variants, data capture, async inference, and shadow variants
Endpoints: Full deployment configuration including blue/green and rolling updates, rollback alarms, and traffic routing policies

Tips and Recommendations¶

The module focuses on provisioning SageMaker. The convention-based approach enables downstream services to easily attach to the SageMaker. Adjust the configuration parameters as needed for your specific use case.

Troubleshooting¶

Reporting Issues¶

If you encounter a bug or issue, please report it on the GitHub repository.

Security Considerations¶

AWS VPC¶

Understand the security considerations related to Sagemaker on AWS when using this module.

Best Practices for AWS Sagemaker¶

Follow best practices to ensure secure Sagemaker configurations:

AWS Sagemaker Security Best Practices

Contributing and Community Support¶

Contributing Guidelines¶

Contribute to the module by following the guidelines outlined in the CONTRIBUTING.md file.

Reporting Bugs and Issues¶

If you find a bug or issue, report it on the GitHub repository.

License¶

License Information¶

This module is licensed under the Apache 2.0 license. Refer to the LICENSE file for more details.

Open Source Contribution¶

Contribute to open source by using and enhancing this module. Your contributions are welcome!