performance-optimization

Fix N+1 queries by replacing per-record lookups with a single query using joins or ORM includes, such as findMany with an include clause. Confirm the fix by checking the database query log before and after the change.

Improve Core Web Vitals by measuring with Lighthouse and the web-vitals library, then targeting the specific bottleneck: optimize LCP images with responsive srcset and fetchpriority, and reduce INP by breaking up long main-thread tasks.

Add an index only after running EXPLAIN ANALYZE shows a sequential scan or stale statistics on the query. Index for the query shape with equality columns first, and skip indexes for low-selectivity filters or leading-wildcard LIKE patterns.

Pool exhaustion shows as all endpoints slowing at once while the database sits idle. Raising the pool size just relocates the queue; instead find what holds connections, and use a proxy like pgbouncer for serverless or autoscaling environments.

Do not optimize without profiling evidence of an actual problem. Premature optimization adds complexity that costs more than the performance it gains, and changes that measure neutral against the baseline should be reverted, not kept.

Prevent regressions by setting budgets for bundle size, API latency, and Lighthouse scores, then enforcing them with bundlesize and Lighthouse CI. Complement synthetic gates with RUM alerting on p75 movements in field data.