We will study the prospects of characterising dark matter at the Large Hadron Collider using novel machine learning techniques. We will constrain ourselves within the framework of simplified dark matter models. If mono-X signals (ie. mono-jet, mono-photon, mono-z, mono-Higgs) are observed, we want to be able to identify the underlying dark matter candidate that can best explain the signal. Using data from all possible signal regions, we want to build an ML model that is capable of inferring these parameters of interest, independent of kinematic cuts, phase space and background contamination. Related compute project: NAISS 2023/22-305 and NAISS 2024/22-277 Related storage project: NAISS 2023/23-411