LLMOps Intermediate — 085: quantify hallucination fallout on `Latency and cache strategy` — memo `603569 [85]`

Lesson 085: Latency and cache strategy

Focus

Anchor this drill to one production LLM workflow—even hypothetical. Token Latency and cache strategy:85 keeps neighbouring lessons differentiable.

Key ideas

• Thread: Latency and cache strategy · drill v5 · spin 25605.
• 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

Eval harness snippet

case_id: LO-14713
route: support_rag_v5
must_include_patterns:
  - "\[chunk_"
forbid_patterns:
  - "guaranteed SLA"
judge_profile: tempered_1

Practice

Practice Draft three eval assertions QA must pass before prompt promotion. — 85 Bump 13.