MadVoro constructs three‑dimensional Voronoi diagrams in distributed‐memory systems using a parallel algorithm built on Delaunay triangulation. The library decomposes a point set across multiple processors, constructs local Delaunay triangulations, and then couples them via distributed communication to form a global Voronoi tessellation. It exposes an API that gives access to cells, faces, vertices, centroids, and cell volumes, enabling detailed mesh analysis at scale. MadVoro supports load balancing and data redistribution across hosts to optimize performance in high‑throughput simulations.