I'm involved into different open source software projects. I also focus on writing good documentation, unit tests and tutorials.
The GSPBox (Graph Signal Processing toolbox) is a software based on spectral graph theory that allows efficient processing of signals living on graphs. It is easy to use, contains many state of the art algorithms and allows quick implementation of your own idea. A python release named Pygsp is also available.
DeepSphere is a spherical convolutional neural network framework based on graph convolutional neural networks. It is freely available and well documented.
The UNLocBoX is a convex optimization software using proximal splitting method that was originally launched with the UNLocX project. The original sofware is written in MATLAB, but a python subproject was born and named the PyUNLocBoX. This software contains a structure to solve convex problems that are very close to their mathematical formulation.
The reproducible research platform is a website where all codes used in my research are posted. As suggested in the name, its purpose is to facilitate any researcher to reproduce my results and to use my algorithms.