This paper presents a novel approach to modeling landslide hazard in a given area by jointly modeling landslide occurrences and their associated areal density over space and time. The study area is the region affected by the 2015 Gorkha earthquake in Nepal, where a multi-temporal landslide inventory was generated using satellite imagery. The model, based on an Ensemble Neural Network (ENN), aggregates landslide data over a 1 km × 1 km squared mapping unit and classifies or regresses it against a nested 30 m lattice. The ENN consists of two components: a landslide susceptibility classifier and a landslide density area regression model. The model is trained using a focal Tversky loss function for the susceptibility component and mean absolute error (MAE) for the density component. The results show promising performance, with an AUC of 0.93 for susceptibility prediction and a Pearson's R of 0.93 for density prediction. The model effectively captures the spatial and temporal dynamics of landslide hazard, providing a comprehensive framework for hazard assessment in a data-driven context. The findings highlight the importance of considering both susceptibility and density in landslide hazard modeling, offering valuable insights for risk reduction and management.This paper presents a novel approach to modeling landslide hazard in a given area by jointly modeling landslide occurrences and their associated areal density over space and time. The study area is the region affected by the 2015 Gorkha earthquake in Nepal, where a multi-temporal landslide inventory was generated using satellite imagery. The model, based on an Ensemble Neural Network (ENN), aggregates landslide data over a 1 km × 1 km squared mapping unit and classifies or regresses it against a nested 30 m lattice. The ENN consists of two components: a landslide susceptibility classifier and a landslide density area regression model. The model is trained using a focal Tversky loss function for the susceptibility component and mean absolute error (MAE) for the density component. The results show promising performance, with an AUC of 0.93 for susceptibility prediction and a Pearson's R of 0.93 for density prediction. The model effectively captures the spatial and temporal dynamics of landslide hazard, providing a comprehensive framework for hazard assessment in a data-driven context. The findings highlight the importance of considering both susceptibility and density in landslide hazard modeling, offering valuable insights for risk reduction and management.