Developed an interactive web platform that models and forecasts wildfire behavior for any selected location. When a user points to an area on the map, the system automatically identifies local data including landcover, elevation, vegetation density, and live meteorological conditions and integrates them into a spatio-temporal LSTM model. The model predicts the likely direction, spread rate, extent, and duration of a potential fire, visualizing results on an intuitive, map-based interface. The platform enables scenario testing by allowing users to modify wind speed, humidity, or fuel conditions, instantly updating forecasts. Designed for use in emergency response, environmental research, and risk assessment, the system transforms complex environmental data into clear, actionable predictions.