As my EE Master's thesis I'll be adding a different way of sharding geospatial data, based on Voronoi diagrams: The user can select a virtual coordinate location for each cluster and geospatial data can then be inserted into the cluster which is "closest" to itself based on the virtual coordinate of the cluster.

This is a more efficient way of sharding geospatial data, since it allows queries to hit fewer servers when searching for data of a certain region and it's also more flexible than the current method used by MongoDB (Quad-Tree) in terms of how it allows the space to be divided among clusters. Source: The attached document, which is an investigation into different geo sharding methods and the reason for my thesis.

I am not at all familiar with MongoDB's codebase and I'm finding it a bit hard to traverse since I can't find any sort of documentation about how the project is organized, if someone could provide me with some pointers it would be very helpful