EngineersofAI LogoCoursesBlogResearch LabAI LettersThe LabCode BankInteractive 3DKodrEarnestJobs
Skip to main content
Interactive 3D/Infrastructure as Code for ML
Infrastructure as Code - Resource Graph
+ 10 applied
~ 0 planned
± 0 changed
- 0 destroyed
aws_vpc
+
ml-platform-vpc
VPC with CIDR 10.0.0.0/16
aws_subnet
+
ml-subnet-az-a
Public subnet AZ-a
depends on: vpc
aws_subnet
+
ml-subnet-az-b
Public subnet AZ-b
depends on: vpc
aws_eks_cluster
+
ml-cluster
EKS v1.29, private endpoint
depends on: subnet_a, subnet_b
aws_eks_node_group
+
cpu-nodes
m5.2xlarge × 3, on-demand
depends on: eks
aws_eks_node_group
+
gpu-nodes
p3.2xlarge × 2, spot
depends on: eks
aws_s3_bucket
+
ml-datasets
Versioned, lifecycle 90d→Glacier
depends on: vpc
aws_s3_bucket
+
ml-model-registry
Model artifact storage
depends on: vpc
aws_ecr_repository
+
ml-training-images
Image scanning enabled
depends on: vpc
aws_db_instance
+
ml-metadata-db
postgres 15.4, db.t3.medium
depends on: subnet_a, subnet_b
# main.tf
resource "aws_eks_cluster" "ml" {
  name     = "ml-cluster"
  role_arn = aws_iam_role.eks.arn
  version  = "1.29"

  vpc_config {
    subnet_ids = [
      aws_subnet.az_a.id,
      aws_subnet.az_b.id,
    ]
  }
}
IaC for ML
Version-control your ML infra
IaC Tool
Plan / Apply
Why IaC for ML
Click-ops infra creates unreproducible environments. IaC gives you git history for your GPU clusters, code review for scale-up decisions, and one-command rollback.

Infrastructure as Code for ML - Interactive Visualization

Infrastructure as Code (IaC) treats cloud resources as version-controlled configuration files rather than manually clicked AWS console buttons. For ML platforms, this means your EKS GPU cluster, S3 model registry, ECR container registry, and RDS metadata database are all defined in Terraform HCL or Pulumi TypeScript. A terraform plan shows exactly what will change before applying - green for additions, yellow for modifications, red for deletions. This makes ML infrastructure reproducible, auditable, and rollback-safe.

  • Explore a complete ML platform resource graph: VPC → subnets → EKS cluster → CPU/GPU node groups → S3 buckets → ECR → RDS
  • Run terraform plan to see all resources in planned state, then apply to converge to desired state
  • Select individual resources to apply, mark as changed, or destroy - see the plan diff update in real time
  • Toggle between Terraform HCL and Pulumi TypeScript syntax for the same infrastructure definition

Part of the EngineersOfAI Interactive 3D - free interactive visualizations covering every major concept in machine learning and AI engineering. Hover any element for a plain-English explanation. No code required.