Walking is a familiar but complex behavior, requiring the continuous coordination of multiple muscles while integrating proprioceptive feedback to correct for unanticipated disturbances. Although a variety of walking models propose mechanisms for this coordination, the neural computations proposed by these models have yet to be experimentally validated. Two crucial barriers are (1) gaps in modeling approaches that integrate specific neural dynamics with realistic kinematics to connect predictions to observations and (2) the inherent difficulty in perturbing or recording from neurons during walking to validate the underlying neural computations. We leverage the powerful genetic tools available in the fruit fly Drosophila melanogaster to address both challenges. To address the first challenge, we propose a feedback model that integrates a model simulating proprioceptor responses, a state estimator to predict angles based on proprioceptor input, and a neural controller to predict fly actions. After training, our model can reproduce the statistics of natural fly joint angles, generate credible walking kinematics, and provide interpretable intermediate variables. To address the second challenge, we develop an experimental framework to validate our model, combining genetically and spatially targeted optogenetic perturbations with kinematics obtained from markerless 3D tracking. We show that the leg joint kinematics of flies following proprioceptive perturbations qualitatively match those predicted by the model. Thus, our approach integrates proprioceptive feedback with a walking control model in an experimentally verifiable way. The advances in the experimental and theoretical frameworks proposed here hold great potential for modeling neural systems with tight feedback loops. In ongoing work, we are extending the model to investigate how proprioceptive feedback is modulated by walking speed and to distinguish hypotheses about coordination among multiple legs. This modeling framework may produce new insight into sensorimotor coordination, with applications in legged robotics and treatment of diseases affecting sensorimotor control.