Realtime SaaS Analytics Dashboard
Replaced a hand-rolled analytics pipeline with a streaming architecture. Query latency p95 dropped 40x. Shipped end-to-end from schema design to dashboard UI.
About this project.
A streaming analytics core for a B2B SaaS platform — built to replace a slow aggregation layer and unlock realtime dashboards for end customers.
Tech Stack
- React + TypeScript Frontend
- Kafka Streaming
- ClickHouse OLAP store
- Node.js + Fastify Backend API
- AWS (EKS, MSK) Infra
Project Info
Key Features
Streaming pipeline
Kafka + ClickHouse ingest handles 80M events per day with sub-300ms p95 dashboard queries.
Realtime UI
React SPA receives streaming chart updates over SSE with saved views and shareable URLs.
Multi-tenant safety
Row-level tenant isolation at the query layer, per-tenant rate limits, full audit log.
Slice + drill
Interactive dashboards with cohort filters, time-range comparisons, and drilldown to row-level events.
Handover
Runbook, on-call playbook, and a trained internal team owning the system long-term.
SLO-driven
Per-route SLO + error-budget dashboards surfaced in Grafana for the on-call rotation.
Feature set
Streaming pipeline
- Kafka + ClickHouse ingest, 80M events/day
- Sub-300ms p95 dashboard queries
- Backfill 18mo history, zero downtime
Realtime UI
- React + TypeScript dashboard SPA
- Streaming chart updates via SSE
- Saved views + shareable URLs
Multi-tenant safety
- Row-level tenant isolation at the query layer
- Per-tenant rate limits + quota enforcement
- Audit log for every dashboard export
Handover
- Runbook + on-call playbook documented
- Team trained on operating model
You might also like.
Open Source: Postgres Migration Diff Tool
Python CLI that diffs PostgreSQL schemas across environments and outputs idempotent migration SQL. MIT-licensed.
View projectCross-Platform Field Service App
Offline-first React Native app for 400+ field engineers. Job tracking, parts inventory, and a sync engine designed for spotty rural…
View project