Evals And Observability easy
A prompt change produced an impressive answer in a live demo for leadership. The PM wants to ship it today. What should you do?
- A Run the change against the existing eval set and compare scores with the current prompt before shipping it.
- B Ship it, because a strong result on a realistic live input is good evidence that the prompt generalizes well.
- C Rerun the same demo input five more times and ship it if the output stays impressive in every single run.
- D Ship it behind a feature flag and skip the offline eval, since online metrics are the real ground truth anyway.
Answer: A. Run the change against the existing eval set and compare scores with the current prompt before shipping it.
LLM output variance is high, so one good demo says little about the many inputs the eval set covers; running the eval delays shipping slightly but catches regressions on other cases. Rerunning the same input only measures sampling variance on one case, not coverage. Online metrics are the ground truth but lag, so skipping the offline gate exposes users to regressions you could have caught in minutes.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- In one release, a RAG assistant got a new embedding model and a new chunking strategy, and the end-to-end judge score dropped 6 points. The team only tracks the end-to-end score and has spent a week tweaking the generation prompt with no recovery. What is the best next step?
- Over two months, your team iterated through 40 prompt versions, each tuned against the same 150-case golden set. The offline score rose from 72 to 91. Online task completion and retry rate have not changed. What is the most likely problem, and what is the fix?
- You plan to replace human graders with an LLM judge for your answer-quality eval. Before trusting its scores in a CI gate, what should you do first?
- After you swap to a smaller model, p95 latency falls from 900 ms to 350 ms and the ops dashboard is all green. Leadership asks whether answer quality is still fine. What is the right response?
- Your CI gate fails any prompt change that drops the score 3 or more points on a 100-point rubric over 60 cases. Production generates at temperature 0.7. Running the eval twice on the same commit gives scores that differ by up to 4 points, and engineers have started rerunning failed gates until they pass. How do you fix the gate?
- Your RAG assistant's answer-quality score dropped from 82 to 70 after a change. You only measure final answer quality. Why is that a problem for debugging?