Abstract: Processes in particle physics are often described by a large number of observables that can carry information on the theory parameters of interest. This proves a challenge for traditional analysis methods, which struggle to extract all of this information. However, recently, a family of new inference techniques combining matrix element information and machine learning has been developed. MadMiner, a Python module wrapping around MadGraph 5 and Pythia 8, automates all steps required for these inference techniques: it supports almost any physics process and model, reducible and irreducible backgrounds, shower effects, detector simulation, and systematic uncertainties, without requiring any approximations on the underlying physics. We demonstrate the use of MadMiner in an example analysis of dimension-six operators.