NEEL RATNPerfDash
An internal performance dashboard turning scattered logs into a single, glanceable view.
Overview
PerfDash surfaces frontend and API performance — page load times, API latencies, error rates — in one place, replacing the manual log-diving engineers were doing after every release.
Problem
No one had a fast way to tell whether a release regressed performance. Debugging meant grepping logs across several services with no shared view.
Architecture & Approach
A lightweight ingestion pipeline feeds a time-series store; the dashboard is built from composable chart primitives with threshold-based alert badges. A query layer lets any team filter by route or service without touching code — the goal was a tool product teams could self-serve, not one that needed me in the loop.
Metrics
Learnings
Internal tools reward ruthless simplicity over feature completeness. Good defaults mattered more than configurability — most teams never touched the filters.