Cloud AI Monitor
Overview
Cloud AI Monitor is a production-grade AIOps platform built during an internship at CIRES Technologies (Tanger Med). It monitors hybrid cloud infrastructure across OpenShift and OpenStack, detects anomalies using a custom mathematical AI model, predicts resource exhaustion, and executes autonomous remediation — all without human intervention.
Architecture
[OpenShift Connector] ──┐
[OpenStack Connector] ──┼──► [PostgreSQL] ──► [AI Decision Engine] ──► [Remediation Engine]
[Grafana Connector] ───┘ │
▼
[React Dashboard + Email Reports]
AI Engine — Anomaly Detection
The core pipeline operates on a three-dimensional anomaly vector:
A = [m, l, t]
Where m = metrics signal (threshold severity, recurrence), l = logs signal (keyword correlation), t = traces signal (latency patterns).
Health scoring uses exponential decay:
H = 100 × exp(-λ × |A|²)
The decision engine maps scores to action tiers: Escalate (≥78), Investigate (≥52), Watch (≥32), Suppress (<32).
Autonomous Remediation
When the AI decides to act, it executes real infrastructure changes:
- OpenShift: Delete stress-test resources, restart crashloop pods, scale CPU/memory limits
- OpenStack: Resize VM flavors, extend volumes, hard reboot or migrate blocked VMs
All actions are logged with full audit trail and email incident reports.
Key Features
- Real-time dashboard with WebSocket updates
- OpenShift pod monitoring (CPU, RAM, restarts, CrashLoopBackOff)
- OpenStack VM monitoring (plug-and-play connector)
- Prediction engine (time-to-exhaustion via linear regression)
- Soft quota system with configurable thresholds
- Stress test / chaos engineering tools
- Cost analysis from real metrics
- Role-based authentication (JWT) with audit trail
- CI/CD auto-deploy on push to OpenShift
Team
- Redouane Meriche — Software & Embedded Systems Engineering, INPT
- Farouk Toujgani — Engineering, EMSI
- Supervisor: Mr. Ibrahim El Hannaoui
CIRES Technologies — Tanger Med | July 2026