Next.js developer for hire

Most slow or broken Next.js apps fail in the same six places. When I take over a project, this is the checklist that runs in the first 48 hours — before any feature work, before any rewrite. It doubles as a free mini-audit if you're curious whether your own build is sound.

1. Bundle splits and client/server boundaries

The first grep is for stray "use client"directives at the top of layout or page files. Mark a layout as client and you've just shipped the entire tree to the browser. I run the bundle analyzer, find the routes pulling Framer Motion or charting libs into the server bundle, and push them behind a dynamic import or a leaf client component.

2. Route segment config and caching boundaries

Next 14+ caches aggressively by default — which means a dashboard that should be dynamic can silently ship stale data, and a marketing page that should be static can hit the database on every request. I audit every route's segment config (revalidate, dynamic, fetchCache) against what it actually does, and alignrevalidateTag calls with the mutations that should invalidate them.

3. Hydration mismatches and double-fetch patterns

“Works on dev, broken on prod” is almost always one of two things: a date or random-value rendered server-side that re-renders differently on the client, or a Server Component fetching the same data its Client child also fetches via TanStack Query. I trace both with the React DevTools profiler and the Next build output, then collapse the duplication.

4. Third-party scripts and the LCP killer

GTM, Intercom, Hotjar, ad pixels — half the perf wins on a typical Next.js site come from auditing what loads, when, and via which Script strategy. afterInteractive belongs on non-critical analytics; beforeInteractive almost never. Most LCP regressions trace back to a marketing-installed script blocking the main thread.

5. Image pipeline and the LCP image

next/image with priority on the LCP image, AVIF/WebP via config, explicit sizes for responsive layouts, and deviceSizes tuned to your actual breakpoints rather than the defaults. A misconfigured image pipeline is usually a one-day fix that buys 600ms of LCP.

6. Edge middleware and auth round-trips

Middleware runs on every request, including static assets unless the matcher is set correctly. I check the matcher, the cookie parsing, and whether auth introspection is hitting an origin database on the edge (which kills latency). The fix is usually JWT verification at the edge with the heavy lookup deferred to the route handler.

Bring me a Next.js project that's slow, hydration-broken, or stuck on Pages Router. The first PR usually lands in week one; the audit lands in 48 hours.