This project is an end-to-end business intelligence pipeline designed to monitor, analyze, and predict mission-critical infrastructure health for a global data centre network. Moving beyond basic reporting, this system simulates an AI-driven predictive maintenance model to flag high-risk assets before they fail, optimizing uptime and reducing reactive maintenance costs.
- Data Generation: Python (
pandas,faker) to simulate 5,000+ realistic hardware incidents, maintenance logs, and telemetry data. - Database Pipeline: Microsoft SQL Server (T-SQL) for relational data warehousing and enforcing a star schema architecture.
- Business Intelligence: Power BI (DAX, Data Modeling) for executive-level visualization and predictive risk scoring.
Key Features:
- Global Threat Mapping: Real-time visibility into revenue bleeding across 6 international facilities.
- SLA Compliance Tracking: DAX-driven calculation to monitor penalty thresholds for downtime exceeding 120 minutes.
- Risk Heatmap: Matrix visualization isolating exact facilities and failure severities.
Key Features:
- Algorithmic Risk Scoring: Custom DAX logic penalizes older assets with a history of emergency (reactive) maintenance.
- Actionable Target List: Prioritized engineering queue showing exactly which assets require immediate intervention.
- ROI Justification: Financial proof demonstrating the cost-savings of predictive maintenance vs. reactive emergency repairs.
To simulate the predictive intelligence, I developed a risk-scoring algorithm prioritizing asset age and historical failure severity:
Asset Risk Score =
VAR BaseRisk = AVERAGE(Asset_Inventory[Age_Years]) * 5
VAR ReactivePenalty = CALCULATE(COUNTROWS(Maintenance), Maintenance[Maintenance_Type] = "Reactive") * 15
RETURN
BaseRisk + ReactivePenalty