The field of machine learning has benefited from rapid advances in recent years, and DeepSig is at the forefront of applying them to signal processing and wireless systems. Building on the foundational work presented at previous GRCons, in this talk we present the technical results and methods from our latest research and experiments, including some previously un-released methods. We will present advances in RF sensing and learned physical layers with autoencoders, new open datasets to be shared with the community, and illustrate fluid interoperability with the GNU Radio and open-source SDR ecosystem. Finally, we will demonstrate several new capabilities and discuss current and ongoing trends in the field.