A company has a model packaged in a container and will integrate it with an existing Python web app. They want to host the model on AWS using Kubernetes, avoid managing the control plane, provision resources reproducibly, and provision the infrastructure using Python. Which solution meets these constraints?

The correct answer is using AWS CDK to provision an Amazon EKS cluster and storing the image in ECR. This solution meets all requirements: EKS provides a managed Kubernetes control plane, avoiding management overhead. AWS CDK allows infrastructure provisioning using familiar programming languages like Python, ensuring reproducibility. ECR is the standard for storing container images on AWS. Using CloudFormation for EC2 instances and setting up Kubernetes manually doesn't avoid control plane management. Using the AWS CLI for EKS provisioning doesn't allow for reproducible infrastructure as code in Python. While CloudFormation can provision EKS, it doesn't meet the requirement of provisioning infrastructure using Python, which CDK directly addresses.