Evals And Observability easy
Your LLM app handles support chats that often contain customer emails and account numbers. Engineers want logs of prompts and outputs for debugging. What is the best default?
- A Log every prompt and output in full, because complete data is the only way to reproduce rare production bugs.
- B Log nothing about prompts or outputs and debug only from aggregate metrics such as latency and error rates.
- C Log a sample of requests with PII redacted before storage, with a short retention period and access controls.
- D Log everything in full but encrypt the log store, since encryption at rest removes the privacy concerns.
Answer: C. Log a sample of requests with PII redacted before storage, with a short retention period and access controls.
Sampled logging with redaction trades some debugging completeness for lower storage cost and privacy risk. Logging everything is tempting for reproducibility, but it stores sensitive data at scale. Encryption at rest does not help here, because anyone with log access still reads the PII. Full logging fits only short-lived, opt-in debug sessions or data that contains no personal information.
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?
- A prompt change produced an impressive answer in a live demo for leadership. The PM wants to ship it today. What should you do?
- 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?