You are an ML engineer at an e-commerce company tasked with building an automated recommendation system that scales during peak shopping seasons. The solution requires provisioning multiple compute resources, including SageMaker for model training, EC2 instances for data preprocessing, and an RDS database for storing user interaction data. You need to automate the deployment and management of these resources, ensuring that the stacks can communicate effectively. The company prioritizes infrastructure as code (IaC) to maintain consistency and scalability across environments. Which approach is the MOST SUITABLE for automating the provisioning of compute resources and ensuring seamless communication between stacks?