InferQuest
← Quest map
Phase 6 · Production

The Serving Surface

Where production bugs actually live: templates, tools, adapters, modalities.

0/6 · 520 XP
You’re previewing the curriculum. Sign in (top right) to track progress, take knowledge checks, and unlock the verifiers.
Tasks
  • Template mismatch is a top source of silent quality regressions. Render a template by hand for one model and diff it against apply_chat_template.

    read+80 XPresource ↗
  • How model-emitted markup becomes OpenAI tool_calls JSON — per-model parsers (hermes/llama/mistral), and where they break. Skim MCP as the emerging layer above.

    read+70 XPresource ↗
  • Deploy one reasoning model on the fleet: separate reasoning_content, cap thinking budgets, and measure what long decodes do to your ITL and cost math.

    build+90 XPresource ↗
  • S-LoRA/LoRAX-lineage batched adapters. Serve ≥2 adapters on one vLLM instance, hot-load a third, and check adapters appear in /v1/models.

    build+120 XPresource ↗
  • Image inputs via content parts (encoder scheduling + image-token budgets), plus a TEI-style embedding server — a different serving profile: no KV cache, latency-critical.

    build+100 XPresource ↗
  • API keys (and what stays unauthenticated: /health, /metrics), prefix-cache timing side channels + cache salting, and a guardrails/prompt-injection primer for the gateway layer.

    read+60 XPresource ↗