Seismic vulnerability modeling is critical to seismic risk assessment, enabling decision-makers to identify and prioritize areas and structures most susceptible to earthquake damage. The use of machine learning (ML) algorithms and Geographic Information Systems (GIS) has surfaced as an encouraging approach for seismic vulnerability modeling due to their ability to integrate and analyze large volumes of data. In this abstract, we present a novel approach to seismic vulnerability modeling that leverages the power of ML and GIS. Using Artificial Neural Networks and Random Forest algorithms, the damage intensity values for an earthquake event with the help of various factors like the location, depth, land cover, distance from major roads, rivers, soil type, population density, and distance from fault lines were predicted. The resulting damage intensity values were classified, keeping the Modified Mercalli Intensity Scale as a reference. The ANN and Random Forest algorithms performed very well in this study, and both the models’ accuracy was above 95% for training and testing data. Utilizing the damage intensity values map, the global seismic hazard map, and other socio-physiological parameters were utilized to generate an exposure grid zonation map. Applying this approach to a case study in the Satara district of Maharashtra highlights the model’s effectiveness in identifying vulnerable buildings and improving seismic risk assessment. This approach provides a valuable tool for disaster management and urban planning decision-makers to develop effective mitigation strategies, prioritize resources, and improve overall disaster resilience.