This project began as a contemplation on Deep Time and sonic imprints of gatherings from various historical times and spaces. These imprints, for example, were various official and unofficial audio archives and audio repositories of historical gatherings, riots and protests. Using unsupervised machine learning, the project intended to sonically imagine gatherings that are yet to come. The machine learning epochs, thus, became spaces where sonic imprints of gatherings from the past could be contemplated upon and gatherings that are yet to come, could be imagined. The process involved de-sonification and subsequently re-sonification of sounds to create new imaginaries and new ways of listening to gatherings of the past that transform our questions of present and future. The project created a range of materials from audio outputs, data visualisations as well as textual outputs that resulted from Natural language processing algorithms specifically designed during the project. The following images are data visualizations that emerged during the training phase of the machine learning algorithm. It embodies the manner in which the algorithm was learning to think through interacting with the data. These data visualizations thus become representative - of the interaction of input data with the architecture of the neural network algorithm.