Ice-penetrating radar data contain a wealth of information about the bed and internal structure of the ice sheet. While these data have long been used to diagnose the presence of basal water or infer attenuation rates, they have rarely been used in a formal inverse model for the ice sheet temperature structure. Here, we invert a coupled thermomechanical ice sheet and basal hydrology model to infer both geothermal flux and accumulation rate from multiple classes of radar observations in the area around Dome A, East Antarctica. Our forward model solves for a coupled steady state between the ice sheet flow field, temperature, and basal hydrology, including melt, water transport, and freeze-on. We fit radar observations of basal water, freeze-on, and internal layers, along with a geothermal flux prior based on aeromagnetic observations (Martos et al., 2017). We minimize the combined misfit function by first using an evolutionary algorithm to find the approximate answer in parameter space, and then optimizing the fit with localized perturbations. In addition to inferring the spatial distribution of geothermal flux and accumulation rate, we are also able to estimate the uncertainty and skewness of their probability distributions, as well as quantify how our result on each individual data constraint. Our results demonstrate a new method for combining multiple glaciological constraints into a single inverse model of the ice sheet, and give us a more rigorous picture of the information content provided by each dataset. In a companion paper we analyze and interpret the best-fit model.