recsys-system-design

Design an industrial recommendation system or subsystem with explicit scale, latency, freshness, consistency, availability, privacy, cost, degradation, observability, and rollout constraints.

SKILL.md

GitHub

Recommendation system design

  1. Resolve objective, QPS, P99, corpus and feature scale, freshness, consistency, runtime, deployment, availability, privacy, and cost. Ask only when alternatives materially differ.
  2. Run python3 tools/recsys_agent.py design "<system>" key=value ... and close all relevant missing constraints.
  3. Search the released PDFs and read the cited pages.
  4. Reuse a runnable template; [MOCK] dependencies are teaching substitutes.
  5. Produce request/data flows, interfaces, capacity and latency budgets, version contract, overload/degradation, metrics, rollout, rollback, and validation.
  6. Separate 本书依据, 工程实现, 验证, and 待业务确认.