A retailer needs a recommendations system that provides recommendations for existing users based on each user's browsing history and filters out items the user already purchased. Which solution meets these requirements with the least development effort?

The correct answer leverages Amazon Personalize, a fully managed machine learning service designed for recommendation systems, minimizing development effort. Specifically, the USERPERSONALIZATION recipe is ideal for generating recommendations for existing users based on their interactions. Creating a real-time filter within Personalize directly addresses the requirement to exclude previously purchased items, which is more efficient than filtering externally. Deploying a Personalize campaign and using GetRecommendations provides a streamlined, real-time recommendation API. The incorrect options involve significantly more development effort. Building and hosting custom collaborative filtering or neural collaborative filtering models in SageMaker requires extensive ML expertise, model training, infrastructure management, and custom API integration (API Gateway, Lambda) for serving and filtering, which contradicts the "least development effort" requirement. While PERSONALIZEDRANKING is a Personalize recipe, it's designed for re-ranking a list of items rather than generating an initial set of recommendations, making USERPERSONALIZATION more appropriate here.