About The Speaker
Luis Contreras is an IT Infrastructure Monitoring Engineer with over 10 years of experience designing and managing enterprise monitoring environments for IT consulting clients across Latin America. Holding both the Nagios Certified Administrator (NCA) and Nagios Certified Professional (NCP) certifications, Luis has spent the better part of a decade helping organizations build reliable, scalable monitoring platforms built on Nagios Core and Nagios XI.
His interest in predictive monitoring grew from a recurring frustration he shared with most of the Nagios Community: being notified after a problem had already impacted users. That frustration led him to explore whether the rich performance data Nagios had been quietly collecting for years could be fed into machine learning models to shift monitoring from reactive to predictive without replacing the platform organizations already trust.
The result is the project he is presenting at NWC 2026: a working, open-source ML sidecar that gives Nagios the ability to forecast threshold breaches up to 30 minutes before they occur, built entirely on top of standard Nagios perfdata.
Luis is based in the Dominican Republic.
From Reactive Alerts to Predictive Intelligence – A Live ML Demo
This session demonstrates how to extend Nagios Core with a machine learning sidecar that transforms it from a reactive alerting platform into a predictive intelligence engine without replacing or modifying Nagios itself.
Using three open-source ML models Facebook Prophet for time-series forecasting, scikit-learn’s Isolation Forest for anomaly detection, and a weighted health score algorithm, the sidecar predicts service threshold breaches up to 30 minutes before they occur, feeds real-time passive check results back into Nagios, and delivers everything through a live dashboard via WebSocket. The result: the Nagios services page turns WARNING → CRITICAL in sync with what the ML model predicted minutes earlier.
Demo
The session includes a fully live demo running on two Debian 12 VMs. The audience watches CPU, disk, and network metrics degrade in real time, sees the ML model predict threshold breaches 2+ minutes before Nagios fires a CRITICAL alert, and observes both the ML dashboard and the Nagios services page respond in sync. The complete lab environment is reproducible from a single Ansible command in under 20 minutes and will be publicly available on GitHub after the conference.
What I Hope You Learn
Every Nagios shop already has the data this system needs, it lives in perfdata files that Nagios has been writing since day one. This session shows that the path from reactive monitoring to predictive intelligence requires no new infrastructure, and no new agents. One configuration line unlocks everything. Attendees leave with a working architecture, a public GitHub repository, and a realistic, no-hype understanding of what ML-powered monitoring actually looks like in practice.
Sign up to get the latest on #NWC2026 speakers, sessions, and registration announcements.
