Complete Terraform for ML Tutorial: Infrastructure as Code for Machine Learning

Terraform is an Infrastructure as Code (IaC) tool that enables you to provision and manage cloud infrastructure declaratively. For ML teams, Terraform helps automate the deployment of training clusters, inference endpoints, and data pipelines.

Why Terraform for ML?

• Reproducibility: Same infrastructure every time
• Version control: Track infrastructure changes
• Multi-cloud: AWS, GCP, Azure support
• Modularity: Reusable infrastructure components
• State management: Track resource state

• ML training infrastructure
• Inference endpoint deployment
• Data pipeline provisioning
• Development environments
• Multi-region deployments

Installation

# macOS
brew tap hashicorp/tap
brew install hashicorp/tap/terraform

Linux
wget -O- https://apt.releases.hashicorp.com/gpg | sudo gpg --dearmor -o /usr/share/keyrings/hashicorp-archive-keyring.gpg
echo "deb [signed-by=/usr/share/keyrings/hashicorp-archive-keyring.gpg] https://apt.releases.hashicorp.com $(lsbrelease -cs) main" | sudo tee /etc/apt/sources.list.d/hashicorp.list

sudo apt update && sudo apt install terraform

Verify installation
terraform --version

ml-infrastructure/
├── main.tf           # Main configuration
├── variables.tf      # Input variables
├── outputs.tf        # Output values
├── providers.tf      # Provider configuration
├── terraform.tfvars  # Variable values
└── modules/
    ├── training/
    ├── inference/
    └── storage/

# providers.tf
terraform {
  requiredproviders {
    aws = {
      source  = "hashicorp/aws"
      version = "~> 5.0"
    }
  }
  requiredversion = ">= 1.0"

}

provider "aws" {
  region = var.awsregion
}

# variables.tf
variable "awsregion" {

  description = "AWS region"
  type        = string
  default     = "us-west-2"
}

variable "environment" {
  description = "Environment name"
  type        = string
  default     = "dev"
}

variable "projectname" {
  description = "Project name"
  type        = string
  default     = "ml-platform"
}

3. Basic Commands

# Initialize Terraform
terraform init

Format configuration
terraform fmt

Validate configuration
terraform validate

Plan changes
terraform plan

Apply changes
terraform apply

Destroy infrastructure
terraform destroy

S3 for ML Data

1. Data Lake Setup

# modules/storage/main.tf
resource "awss3bucket" "mldata" {

  bucket = "${var.projectname}-${var.environment}-ml-data"

  tags = {
    Environment = var.environment
    Project     = var.projectname

    Purpose     = "ML Data Lake"
  }
}

resource "awss3bucketversioning" "mldata" {

  bucket = awss3bucket.mldata.id
  versioningconfiguration {

    status = "Enabled"
  }
}

resource "awss3bucketlifecycleconfiguration" "mldata" {
  bucket = awss3bucket.mldata.id


  rule {
    id     = "archive-old-data"
    status = "Enabled"

    transition {
      days          = 90
      storageclass = "STANDARDIA"

    }

    transition {
      days          = 180
      storageclass = "GLACIER"
    }
  }
}

Folder structure
resource "awss3object" "rawdata" {

  bucket = awss3bucket.mldata.id
  key    = "raw/"
  contenttype = "application/x-directory"

}

resource "awss3object" "processeddata" {
  bucket = awss3bucket.mldata.id

  key    = "processed/"
  contenttype = "application/x-directory"
}

resource "awss3object" "models" {