While all of Earth’s land surfaces have been mapped at 30 m resolution or finer, the topography of the seabed is still largely a black box, with only 26.1% of the seabed mapped to “adequate resolution” thus far. In practice, this means that compiling a gridded bathymetric dataset often requires “filling in” areas of missing data, “smoothing” conflicts between overlapping datasets, and “blending” information from multiple sources. This talk discusses several strategies for fusing and interpolating datasets such as coastal lidar, high resolution multibeam echosounder data, chart depths, and the globally available GEBCO grid in real world application. The “fuzzy” boundary between land and sea is emphasized, as terrestrial and marine datasets often disagree when their coverage overlaps. Interpolation and smoothing methods are explored as well. The techniques discussed are implemented in Python using geospatial libraries and thus open source, easily scaled, and reproducible.