Seismic waves are vibrations caused by energy from earthquakes. Over 58.6% of the land in India is susceptible to earthquakes and over 90% of casualties in India were caused by building collapses. To protect lives and property, seismic risk prediction is done by calibrating grid-wise risk for the city. As manual building-by-building risk evaluation is impractical, automatic seismic risk prediction using machine learning is its solution. It involves clustering seismic zones based on development trends tied to habitation and income groups, discerned through satellite image analysis. Features like street intersection density, tree density, built-up areas, demographics, important structures (e.g., hospitals, restaurants, tourist attractions, etc.), and building age are extracted from satellite images using GIS tools. Machine Learning techniques, including Random Forest Classifiers and Neural Networks, are then applied to determine feature relevance and to obtain seismic risk zones. This leads to vulnerability assessment, estimating economic and life losses. The proposed GIS and ML-based approach streamlines feature extraction, and zone (100-m grid) classification with an approximate accuracy of 74% simplifying seismic risk assessment.