Real-time systemsCommercial fleet logistics
Real-Time Fleet Intelligence Platform
Client and product name withheld under NDA
A platform that ingests continuous vehicle telemetry, keeps live trip state, detects abnormal behaviour, and answers operator and driver questions using the same operational data it monitors. The AI layer is not a detached chatbot: it reads current vehicle context, resolves nearby locations through function calling, and streams answers back while trips are still in progress.
RoleBackend architecture, real-time pipeline, AI interaction model, monitoring frontend
Capabilities
- Real-time vehicle tracking
- Trip and route visualization
- Driver-facing AI copilot
- Function calling for nearby locations
- Weather, traffic and vehicle context integration
- Telemetry anomaly detection
- Streaming AI responses
- Trip and conversation history
Challenges
- Absorbing frequent telemetry events without falling behind
- Keeping the latest vehicle state cheap to read
- Running geospatial searches inside request budgets
- Streaming model responses reliably across reconnects
- Grounding AI answers in live operational data
Technologies
- NestJS
- TypeScript
- React
- PostgreSQL
- PostGIS
- Redis
- RabbitMQ
- Server-Sent Events
- WebSockets
- AWS
- Google Maps
- OpenAI