AI Analytics & AI Creative Insights
Data-heavy analytics features at Entropik Tech, surfacing facial-coding and eye-tracking results through responsive React interfaces.
+20%
Feature adoption
+18%
Search conversion
+40%
UI responsiveness
Problem
Rendering heavy facial-coding and eye-tracking analytics on the main thread blocked the UI and hurt perceived performance for enterprise clients.
Approach
Moved SDK processing off the main thread and built a virtualized search interface that cancels redundant requests.
Architecture
- Web Worker-based offloading of heavy SDK computation.
- IndexedDB for client-side caching of processed analytics data.
- React virtualization for large result sets.
Technical challenges
- Heavy SDK computation froze the UI for enterprise users running these workloads in the browser.
- Large result sets and fast typing caused slow rendering and redundant API calls.
Performance decisions
- Offload SDK processing to Web Workers so input and scrolling stay responsive.
- Virtualize result lists so render cost no longer scales with list size.
- Cancel superseded search requests with AbortController.
- Reduced bundle size by 30% through code splitting and import-graph analysis, improving Time to Interactive by 600ms.
Stack
- React
- TypeScript
- Redux Toolkit
- Web Workers
- IndexedDB