Monitoring and operations automation

Context

Daily operations included repetitive tasks and manual information handling that could be standardised.

Responsibility

Design and evolution of automation around Zabbix, alerts, analysis and procedures, using AI as technical support where it adds value.

Architecture

Small controlled workflows, scripts and validations connected to operational events, with human review for sensitive actions.

Implementation

Python, Bash and PowerShell scripts; input normalisation; generation of diagnostic context; and workflow documentation.

Challenges

Avoiding opaque automation, controlling errors and retaining traceability when a task is no longer manual.

Outcomes and lessons

More repeatable processes, better-prepared diagnostic information and less effort spent on mechanical tasks.

Automation does not mean removing judgement: it means preserving it for the decisions where it adds real value.