LLMOps Basic — 099: pair eval regressions on `Staging deploy realism` — memo `539398 [99]`

Lesson 099: Staging deploy realism

Focus

Prefer explicit failure rehearsals over aspirational wording. Token Staging deploy realism:99 keeps neighbouring lessons differentiable.

Key ideas

• Thread: Staging deploy realism · drill v9 · spin 757714.
• Habit: attach a trace_id to every completion you would paste into an ops dashboard.
• Guardrail: add one RACI bullet for prompt or index changes before tomorrow's standup.

Deep dive notebook

Synthetic drill artefacts

Tool contract rehearsal

{
  "name": "reindexSlice",
  "arguments": { "corpus": "corpus-9", "dry_run": true }
}

Preconditions: index lag < 20h
FAILURE: page platform if lag > SLA without VP memo.

Practice

Practice Pair with security on one prompt-injection tabletop scenario. — 99 Bump 34.