The problem
A Gulf-based logistics company with 1,200+ vehicles had GPS trackers on every truck — but the data was sitting in a raw database that nobody could interpret. Fuel consumption was 18% above industry benchmark. Maintenance was reactive, not predictive. Route inefficiencies were invisible until they showed up as customer complaints.
They had the data. They didn't have intelligence.
What we built
1
Real-time data ingestion pipeline
GPS telemetry from 1,200+ vehicles streams via MQTT broker into a Django backend. Data normalized, enriched with route metadata, and stored in TimescaleDB (PostgreSQL time-series extension). Processing latency under 2 seconds end-to-end.
2
AI anomaly detection engine
An Isolation Forest model trained on 18 months of historical telemetry detects fuel consumption anomalies, unusual idle patterns, route deviations, and engine behaviour outliers. Anomalies trigger configurable alerts — push notification, email, or dashboard flag.
3
Predictive maintenance scoring
A regression model trained on historical breakdown data predicts maintenance probability for each vehicle over the next 7, 14, and 30 days based on mileage, engine hours, temperature readings, and brake usage patterns. Fleet managers see a ranked risk list each morning.
4
React dashboard with live WebSocket feed
A React frontend consumes live vehicle positions, anomaly alerts, and AI insights via WebSocket connection. Map view (Mapbox GL), vehicle list with status, anomaly timeline, and fuel report generator. Fully responsive — works on the operations center wall screen and on a tablet in the field.
5
AI-generated route insights
An LLM analyses completed routes and generates natural language summaries highlighting idle time, traffic holds, and concrete suggestions such as shifting departure windows or rerouting via faster corridors.
We had the GPS data for years. We never knew what to do with it. Now the system tells us exactly which trucks to service this week, which drivers are burning excess fuel, and why. We saved more in fuel in the first quarter than the platform cost us.
— Head of Fleet Operations, Logistics Company, UAE
