Modern radio telescopes have seen tremendous gains in sensitivity, field of view, and frequency bandwidth, thanks largely to advances in signal processing. This has allowed us to uncover entirely new source classes, such as fast radio bursts (FRBs). Beyond the mystery of their origin, FRBs have held promise as a new cosmological probe ever since their discovery in 2007. My talk will lie at the intersection of FRB cosmology, large radio surveys, and astroinformatics. I will first discuss applications to which FRBs are uniquely suited, focussing on the “missing baryon problem” and coherent gravitational lensing. I will argue that mapping the Universe’s baryons with FRBs will make invaluable contributions to our understanding of feedback, galaxy formation, and weak lensing systematics once we have tens of thousands of localized FRBs. In the second chapter of my talk, I will introduce the DSA-2000, a powerful new telescope that will lead to a quantum leap in radio survey science when it is deployed in 2026. Combined with the Canadian telescope CHORD, we could have a sample of 100,000 localized FRBs by the end of this decade. Finally, I will show how deep learning-based algorithms will accelerate several of the science cases described in my talk and I will highlight recent work I have done using computer vision for interferometric imaging.