Atmospheric reentry is characterized by a combination of complex physico-chemical phenomena including thermal and chemical non-equilibrium. Predictive simulations using detailed aerothermodynamics models are essential in order to study these phenomena and estimate the generated heat flux during the descent phase. These calculations provide verification of experimental campaigns in ground-testing facilities and are key to design flight experiments. The numerical simulation of non-equilibrium plasma flows surrounding spacecraft during reentry conditions is a challenging problem because of several aspects: chemical and thermal non-equilibrium, the multi-scale character of the flow and the large number of species and reactions involved. At BRITE we use data-driven and machine-learning techniques to reduce the computational time of these simulations.