Backend Performance Audit

Details

We profiled every GraphQL path serving workspace dashboards, chat, and activity, then fixed the top thirty N+1 queries with select_related and prefetch_related. Cache keys were tightened to include tenant scope. Under the same load, median response times dropped meaningfully and the tail (p95) compressed. Nothing about your product changed, it just feels faster.

What changed

• Top thirty N+1 query patterns eliminated
• Tenant-scoped cache keys across every cached resolver
• Median and p95 response times improved under load test