Whenever you're ready, walk me through how you'd approach this.
Design and Implement full system solution
Describe and implement a complete production machine learning system from data ingestion through training, model serving, evaluation, and monitoring. Treat the task as open-ended: clarify the prediction objective, users and traffic, data availability, latency and reliability requirements, and whether retrieval, ranking, or re-ranking stages are needed. Provide an executable implementation plan, interfaces between components, offline and online evaluation, and handling for drift, training-serving skew, failures, and model rollback.