Krypton
- Industry
- Developer Analytics
- Team
- 2 engineers
- Project outline
- Ship a self-hosted analytics engine that answers product questions in milliseconds while collecting no personal data — installable with a single Docker command.
The Challenge
Privacy-friendly analytics usually means aggregated, shallow data — and self-hosted tools tend to demand a database administrator. Krypton had to be deep enough for real product decisions and simple enough to run on a five-dollar VPS.
The Solution
Events are ingested by a small Go service that strips identifying data at the edge — no cookies, no fingerprinting — and batches writes into ClickHouse, whose columnar storage keeps p99 query latency in single-digit milliseconds over hundreds of millions of rows.
The dashboard is a React app served by the same binary. One container, one port, zero configuration files.
Outcome
A production instance handles ~40M events per month on 2GB of RAM. The zero-config promise held: median time from docker run to first dashboard is under a minute.