// automate

Azure FinOps and cost anomaly detection

A cost reporting pipeline on Azure Cost Management exports, with a daily anomaly-scoring model that catches resources left running unnecessarily.

Role
Engineer
Status
In production

Python · Azure Automation · Azure

Problem

Cost reporting was a manual process.

What it does

  • Monthly report. Compares against prior months and breaks cost down by cost center, resource group, subscription and resource type, as a formatted Excel report with charts.
  • Daily month-to-date report. Trend, end-of-month projection, top resource groups and resources, and a summary per cost center posted to Teams.
  • Anomaly scoring per resource group, flagged at a score of 30 or more, from rules such as new spend after a zero baseline, spikes of at least 2x the average and sustained increases.
  • An idempotent anomaly log that the monthly report reuses.

Result

  • Replaces a manual monthly process and flags resources left running unnecessarily every day.

← Selected work