An AI Engineer is experimenting with data retrieval performance within a RAG system.

Query expansion with clarifying keywords and synonyms improves recall without abandoning relevance. A single keyword is usually too brittle for semantic retrieval. The durable control mechanism is a separated data plane where ingestion, indexing, retrieval, reranking, and generation can each be measured and updated. The selected option specifically A states “Adding clarifying keywords and synonyms to the original query to broaden the search.”, which matches the operational requirement rather than a superficial wording match. Option A fits the operating model because the problem describes an agent that must remain adaptive under changing inputs and infrastructure conditions. The alternatives would look simpler in a prototype, but synchronous monoliths make freshness and latency fight each other because indexing and generation cannot scale independently. This lines up with NVIDIA guidance because a production RAG workflow should treat the retriever as a measurable service, not as an invisible prelude to LLM generation. For certification purposes, read the question as asking for controlled autonomy, not raw LLM creativity.