Design or review an end-to-end ML/AI system across data, training, serving, MLOps, and scale — including feature stores, model serving, evaluation, and feedback loops. Use when the user asks to "design a recommendation/ranking system", review an ML pipeline, or discuss drift, model serving, or evaluation. For RAG specifically use rag-system-design; for LLM apps use llm-app-architecture; for generic (non-ML) systems use architecture-review.
