Building Analytics UI That Scales to Billions of Events
Rollups, funnel charts, retention cohorts, and the UX patterns that keep exploration fast even on mock data.

Analytics products fail in the UI long before they fail in the warehouse. If every slice requires a spinning loader, analysts stop asking questions. The interface has to feel instant even when the backend is doing heavy aggregation.
Precompute what humans repeat
KPI tiles, hourly throughput, channel breakdowns, and retention cohorts are not ad-hoc queries — they are curated views with known shapes. Mock APIs can still mirror that contract so the frontend learns the real constraints early.
- Typed overview payloads validated with Zod
- Metric toggles that reuse the same chart shell
- Funnel and cohort components isolated for testing
- Shareable state encoded in the URL where possible
Even in a portfolio demo, treating analytics like a product surface makes the engineering story credible.