This paper introduces an iterative neural network-based approach to solve finite-horizon continuous-time stochastic control problems with jumps, when the underlying dynamics are fully known and given.