Abstract: Classical simulations of quantum many-body systems are essential for both studying and understanding quantum technologies. In recent years, with the improvement and the development of several methods such as tensor networks and neural quantum states, classical simulations can reach outstanding results in both efficiency and accuracy. In this framework, we present a novel approach based on deep learning density functionals for quantum many-body systems. density functional theory can bypass the wavefunction approach for solving the many-body problem increasing the speed up and with good accuracies. By using deep learning, density functional can be easily provided without approximations and with adaptive functional forms. Here, we present the idea based on the deep learning density functional theory applied to continuous systems, with its limits and novel strategies to overcome them. We also show how the theory and the deep learning method can be extended to spin systems. Finally, we show how it is possible to extend the density functional for time dependent spin models and the application of DL-density functionals into that framework.