🌍 ClimateTwin AI — AI-Powered Environmental Intelligence

ClimateTwin AI is an environmental intelligence platform that transforms raw environmental observations into detection, explanation, prediction, and actionable insight.

The project was inspired by a simple gap: environmental sensors can generate large volumes of measurements, but raw data does not immediately tell us what is unusual, why it matters, or what could happen next.

ClimateTwin addresses this by building an end-to-end pipeline around real JKUAT Conduit environmental data. The current dataset contains 2,089 observations covering variables such as temperature, humidity, wind, heat index, pressure, UV, and wet-bulb measurements.

Instead of building a static dashboard, we combined multiple intelligence layers:

Conduit Data → Data Processing → Anomaly Detection → Environmental Event Intelligence → Short-Term Forecasting → What-If Analysis → RAG + Gemini Copilot

The platform can identify unusual multivariable observations, classify them into environmental events, assign priorities, generate recommended actions, forecast environmental conditions for the next 1–3 hours, and allow users to test hypothetical scenarios.

The ClimateTwin Copilot connects these outputs with a retrieval-augmented knowledge layer, allowing users to investigate observations, anomalies, forecasts, concepts, and What-If scenarios using natural language.

One of our biggest goals was to make the system transparent and responsible. Predictions are clearly presented as model outputs, What-If results are treated as hypothetical scenarios, and anomaly detection is not presented as proof of a dangerous real-world hazard.

ClimateTwin AI turns environmental data → intelligence → prediction → explanation → impact.