Skip to content
ML Academy

RAG and agents

Interview question

  • stafflevel
  • 30 minsuggested time
  • 2Applied AI, ML engineer

Five product teams are building RAG features on a shared retrieval platform. Documents change daily, teams change prompts weekly, and the model provider updates models a few times a year. Design the evaluation and release process so quality does not silently degrade.

Tests: LLMOps at platform scope: versioning of data, prompts, retrievers and models; golden sets; regression gates; tracing; ownership; accepted-regression records.

day 43day 45day 51day 52day 55

Sign in to practise: answers are saved before model answers are shown.