SageMaker with Terraform

Amazon SageMaker is the new service in AWS to build, train, and deploy machine learning models at scale.

We've already started to build Terraform support for SageMaker and if you don't want to wait until its released, we have good news - you basically can start using it right now. Below, we show how.

You find the Terraform code in main.tf.

Prerequisites

• awscli (>= 1.14.2)
• terraform (>= 0.10.0)

Build a custom provider

As SageMaker support isn't released yet, you need to build your own custom AWS provider based on the new feature branch:

    git clone https://github.com/cloudetc/terraform-provider-aws.git \
        $GOPATH/src/github.com/terraform-providers/terraform-provider-aws

    cd $GOPATH/src/github.com/terraform-providers/terraform-provider-aws/
    git checkout feature/resource-aws-sagemaker-endpoint

    make build

    cp $GOPATH/bin/terraform-provider-aws  ~/.terraform.d/plugins/

Deploy an example model

Let's use the example provided by AWS to show how to deploy your own algorithm container with SageMaker using Terraform:

    # clone the example repo from AWS
    git clone https://github.com/awslabs/amazon-sagemaker-examples.git
    cd amazon-sagemaker-examples/advanced_functionality/scikit_bring_your_own/container/

    # export credentials for your account
    export AWS_ACCESS_KEY_ID=<your-access-key-id>
    export AWS_SECRET_ACCESS_KEY=<your-secret-access-key>

    # create docker algorithm container and push it to ECS
    ./build_and_push.sh foo

    git clone https://github.com/cloudetc/terraform-sagemaker-example.git
    cd terraform-sagemaker-example/
    terraform init
    terraform apply

    # make test call to the deployed model
    aws runtime.sagemaker invoke-endpoint --endpoint-name terraform-sagemaker-example \
        --body "`cat payload.csv`" --content-type "text/csv" "output.dat"

    # show the predicted values
    cat output.dat

    # destroy the deployed model
    terraform destroy