Elucidating the wiring diagram of the human cell is one of the central goals of the post-genomic era. Here, we integrate genome engineering, confocal imaging, mass spectrometry and data science to systematically map protein localization in live cells and protein interactions under endogenous expression conditions. For this, we generated a library of 1,311 CRISPR-edited cell lines harboring fluorescent tags that also serve as handles for affinity capture, and applied a new machine learning framework to encode the interaction and localization profiles of each protein. Our approach provides a data-driven description of the molecular and spatial networks that organize the human proteome. We show that unsupervised clustering of these networks delineates functional groups and facilitates biological discovery, while hierarchical analyses uncover the core features that template cellular architecture. Furthermore, we discover that localization signatures are remarkably predictive of protein function, and often contain enough information to identify molecular interactions. Paired with a fully interactive website (opencell.czbiohub.org), OpenCell is a resource for the quantitative cartography of human cellular organization.