AI Fire Spread Prediction Platform

I need an end-to-end, AI-driven planning and orchestration environment that ingests real-time fire sensor readings from land-based assets, fuses them with topographic layers and live weather feeds (wind speed, humidity, temperature), and continually predicts how the fire will propagate. The core objective is accurate, near-real-time forecasting of the fire’s spread so responders can position resources and plan evacuations before conditions change.

Here’s what matters most to me:
• A robust data pipeline able to stream and reconcile varied sources (IoT, AIS, satellite, SIG weather, GIS shape files).
• A predictive model—deep learning, physics-based, or hybrid—that outputs rapid, high-resolution spread maps plus confidence scores.
• A lightweight orchestration layer or dashboard that visualises the forecast, lets operators simulate containment tactics, and triggers automated alerts via standard APIs (REST/MQTT/Kafka).
• Deployment flexibility: containerised services ready for an on-prem shipboard server as well as a cloud instance.

Acceptance criteria
1. Demonstrate live ingestion of at least one sensor feed, a topographic dataset and a wind feed, then generate a 30-minute forecast , only forecast will do
2. Visualise the predicted perimeter on an interactive map with time-scrub capability.
3. Provide clear documentation: data schema, model architecture, API endpoints, and a repeatable deployment script (Docker-Compose/Kubernetes).
4. Accuracy benchmark: only prototype

If you have experience combining geospatial analytics, IoT streaming and machine-learning prediction, I would love to see how you could bring this critical safety platform to life.