Numerical Weather Prediction systems, while fundamental in meteorology, face critical limitations in computational costs, speed, resolution, and the ability to predict extreme weather events. This thesis explores the potential of deep generative modeling to address these challenges, with a focus on precipitation nowcasting using CASA radar data. Furthermore, the thesis introduces a prototype of a web platform that streamlines access to CASA’s atmospheric data, aiming to simplify workflows for researchers such as identifying and analyzing weather events and effectively accessing data to train machine learning models. This dual emphasis on innovative modeling and enhanced data accessibility seeks to advance the field of nowcasting and provide valuable tools for the atmospheric research community.