The three-dimensional organization of the genome affects cell function and can be interrogated in an unbiased manner via methods such as Hi-C, a genome-wide proximity ligation assay. Hi-C assays have led to a deeper understanding of the mechanisms underlying chromosome conformation, such as chromatin loops and compartments. Improvements to the Hi-C protocol have made it possible to generate meaningful contact maps down to base-pair-resolutions, comparable to the types of resolutions used to analyze most epigenetics assays, such as ChIP-Seq. Generating these ultra-high-resolution maps, however, requires terabases of DNA sequencing. We developed an ecosystem of open-source software tools to support the analysis and visualization of ultra-deep Hi-C datasets. These tools include the Juicer 2.0 pipeline for processing billions of DNA sequencing reads into contact maps at base-pair resolutions; the straw library for powering rapid programmatic access to Hi-C data from both local and remote files; a novel ensemble deep learning approach to annotate chromatin loops; and a novel algorithm that combines dimensionality reduction with unsupervised learning to reliably identify genomic subcompartments. Together, these tools enable the comprehensive processing and analysis of Hi-C data to unprecedented resolutions. We then applied these tools to generate and fully annotate 10-base-pair-resolution maps of nuclear architecture in over 100 primary samples and cell lines, spanning over 40 unique human tissues. Taken together, these datasets report over 100 terabases of raw sequence data, revealing hundreds of thousands of DNA loops localized down to 10-base-pair resolution, and help elucidate the function of genome architecture across human anatomy.