Béranger Thomas
Go TypeScript Monitoring VictoriaMetrics Time Series Machine Learning Docker

Veona

System monitoring platform: lightweight Go agent, TypeScript/Hono server, VictoriaMetrics time-series storage and integrated ML engine (anomalies, forecasting, health scoring).

Context

Veona is a system monitoring platform designed and implemented from the collection agent to the visualisation layer. The goal: a lightweight, reproducible stack for collecting system metrics, storing them as time-series data, and detecting anomalies — without relying on any third-party cloud service.

Architecture

The platform is organised into three separated planes:

Key components

Features

Technical notes

The project uses standard statistical algorithms (Z-score, linear regression) where heavier models would be disproportionate: inference latency stays below one millisecond, with no dependency on an external ML runtime.

GitHub ↗