Tagged performance

Measure page speed, API latency and build times against stored baselines

Find the bottleneck in slow ETL, backfill or export jobs and speed them up safely

Cut p95 and p99 latency in realtime feeds, queues and caches with live readbacks

Detect web performance regressions by benchmarking pages before and after changes

Review LLM cost, latency, errors and routing evidence to see where spend goes

Find and fix performance bottlenecks across frontend, backend and database queries

Monitor provider health, circuit breakers, latency percentiles and budget alerts

Track benchmark drift in win rate, latency and cost across saved runs

Set cache headers so browsers reuse static files instead of downloading them again

Run browser based performance audits with Lighthouse, DevTools or field data

Serve static assets from a CDN to cut latency for users far from your server

Turn on gzip or Brotli compression so text assets transfer far fewer bytes

Set up Google Consent Mode v2 so tags respect visitor cookie choices

Shorten critical request chains so the page renders before extra files load

Shrink CSS file size by removing unused rules and dead styles

Reduce cumulative layout shift so content does not jump while loading

Reduce DOM node count and nesting so pages render and update faster

Remove duplicate JavaScript libraries that load the same code twice

Speed up first contentful paint so visitors see content sooner

Load web fonts efficiently to avoid invisible or shifting text

Optimize Google Tag Manager loading so tags do not slow page speed

Cut the number of HTTP requests a page makes on load

Enable HTTP/2 or HTTP/3 so many assets load over one connection

Make page interactions respond faster by reducing work on each input

Shrink JavaScript bundles by splitting and dropping heavy dependencies

Use secure, up to date JavaScript libraries and replace risky old versions

Speed up largest contentful paint so the main content appears quickly

Turn off lazy loading for above the fold images so they load immediately

Lazy load offscreen images and iframes to cut initial page weight

Stop shipping legacy JavaScript polyfills to modern browsers

Virtualize long lists and tables only when DOM cost is the real bottleneck

Show loading indicators so users know when content is still on its way

Keep page load time under 3 seconds and check it with real measurements

Keep total page weight under 1500 KB by trimming assets and scripts

Set bundle size and Core Web Vitals budgets that CI enforces on every change

Use resource hints to start key downloads before the browser finds them

Preconnect to critical third party origins to shave connection setup time

Add resource hints so the browser fetches important files earlier

Publish source maps for production builds so errors trace to real code

Load third party scripts without blocking the page or hurting speed

Analyze page performance with WebPageTest waterfalls and filmstrips

Keep KV cache prefixes stable to cut agent latency and token cost

Score the reasoning and writing patterns in a sample text with an exploratory rubric

Profile and optimize PixiJS v8 apps for frame rate, draw calls and GPU memory

Separate read and write models with CQRS for scalable query performance

Track down bugs and slow code with systematic debugging and profiling tools

Trace requests across microservices with Jaeger and Tempo to find bottlenecks

Profile slow Python with cProfile and memory tools, then optimize bottlenecks

Add tracing, metrics and SLOs to service mesh traffic for latency debugging