Redis Caching Layer - Sub-200ms query responses

[jordanpartridge]

Goal

Achieve sub-200ms query responses through aggressive Redis caching

Why

Read-optimized architecture for instant AI context injection

Tasks

• Cache embeddings (key: hash of text, TTL: 7 days)
• Cache search results (key: query hash, TTL: 1 hour)
• Cache Qdrant collection stats (TTL: 5 minutes)
• Implement cache warming on startup
• Add cache invalidation on entry updates
• Monitor cache hit rates in know stats

Success Criteria

• Cached query < 50ms (90th percentile)
• Uncached query < 200ms (90th percentile)
• 80%+ cache hit rate in normal usage
• know stats shows cache metrics

Priority

🟡 HIGH - Performance critical for AI integration

Related

Part of strategic pivot to AI-first semantic context engine. See ROADMAP.md