You have the choice of deployment via CDK (recommended) or manual deployment.

Find our example CDK stack here: https://github.com/DGG3D/marketplace-deployment-example 

In case you want to deploy manually:

1. Subscribe to RapidCompact https://aws.amazon.com/marketplace/pp/prodview-zdg4blxeviyyi 
2. Create S3 Buckets: Create one S3 bucket to store your input models and another S3 bucket to store your output models. Ressource on how to create S3 buckets: https://docs.aws.amazon.com/AmazonS3/latest/userguide/create-bucket-overview.html 
3. Create Fargate Cluster: Open the ECS console, create a new AWS Fargate (serverless) cluster.
4. Create a new Policy: Go to IAM and create a new Policy (JSON) with the following content (make sure to adjust INPUT-BUCKET and OUTPUT-BUCKET with the respective bucket names): https://raw.githubusercontent.com/DGG3D/marketplace-deployment-example/main/policy.json 
5. Create Task Role: Create a new IAM Role and select the previously created policy
6. Task Definition: Create a new task definition
   • Provide a name
   • Select launch type: AWS Fargate
   • Select OS: Linux/X86_64
   • Choose Task size based on your model requirements: For simple models 1vCPU + 3 GB RAM is enough, for more complex models you might want to increase this.
   • Task Role: Select the IAM role you created in step 4
   • Container: Use 709825985650.dkr.ecr.us-east-1.amazonaws.com/darmstadt-graphics-group-gmbh/rapidcompact-renewal:0.0.4 as Image URI
   • Enable "Use log collection" and select Amazon CloudWatch
   • Navigate to your newly created Task Definition and note the container name for running the task later
7. Choose a public subnet for your task to run in
   • Navigate to AWS VPC, there should already be a default VPC, if you prefer to use a different VPC note it's ID
   • In the sidebar navigate to 'Subnets', choose a subnet that is in your VPC and note it's ID as well (you can also create a new subnet here if necessary, resource on VPC Subnets: https://docs.aws.amazon.com/vpc/latest/userguide/create-subnets.html )
8. Choose or create a security group
   • Navigate to AWS EC2 -> Security Groups
   • Choose a security group or create a new one and note it's ID (it can be as restrictive as required, this will not affect functionality)
9. Run Task through AWS CLI:
   • Install AWS CLI and authenticate it using your account's credentials: https://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html 
   • Make sure your input model is uploaded to your input bucket
   • Then adjust the following command based on your CLUSTER_NAME (of the cluster we created), TASK_DEFINITION_NAME (name of the task definition we created in step 5), CONTAINER_NAME (the name of the container to run the task on, see step 5), SUBNET (the ID of the subnet to run the task in, see step 6), SECURITYGROUP (the security group to run the task with, see step 7), INPUT_BUCKET, OUTPUT_BUCKET, INPUT_FILENAME (make sure your input bucket contains your input file with this name), OUTPUT_FILENAME (you can choose this freely as it will be created on the output bucket):

aws ecs run-task
--cluster CLUSTER_NAME
--task-definition TASK_DEFINITION_NAME
--launch-type FARGATE
--network-configuration "awsvpcConfiguration={subnets=[SUBNET],securityGroups=[SECURITYGROUP],assignPublicIp=ENABLED}"
--overrides '{"containerOverrides":[{"name":"CONTAINER_NAME", "command":["/bin/sh", "-c", "aws s3 cp s3://INPUT-BUCKET/INPUT_FILENAME . && /rpdx/rpdx -i INPUT_FILENAME -e OUTPUT_FILENAME && aws s3 cp OUTPUT_FILENAME s3://OUTPUT_BUCKET/OUTPUT_FILENAME"]}]}' ``` 10. Monitor and Support: Monitor task execution with CloudWatch logs and reach out for support via AWS Marketplace. 11. More Examples: For an example CDK stack with the proposed setup configured see here: https://github.com/DGG3D/marketplace-deployment-example