The principals of QGNai are organized around development of a novel machine learning platform (designated Categorifier) that is designed and intended significantly to improve the learning and decision-making of machines. Having broad applications in natural language processing, genomic research, drug discovery, cybersecurity, and robotics, Categorifier learns hidden relationships in the data - this might be an image, text, bio sequence, etc. - through a hierarchical structure resulting in deeper learning than state of the art. ? Using a proprietary recursive dimensional reduction technology, the platform coarse-grain initial datasets to discover their underlying dominant features at all levels. The invertibility of the architecture allows fine-tuning of coarse-grain data to generate new datasets based on latent variables. Categorifier's bidirectional generative design enables efficient classification, clustering, and explainability.