RAG (Retrieval-Augmented Generation)

In recruiter context

RAG is the dominant pattern for production AI features in 2026 because it solves two LLM problems at once: hallucination (retrieved context grounds the model in real data) and freshness (the model can answer questions about data that didn't exist when it was trained). The architecture layers usually run: ingestion → embedding → vector storage → retrieval → reranking → generation. AI Engineer interviews in 2026 frequently ask candidates to architect a RAG system at scale — chunking strategy, embedding model selection, retrieval evaluation, the works. The Achilles heel of RAG is retrieval quality: if the wrong context is fetched, the answer is wrong even with a perfect model.