Intelligent recommendation applications in smart cities require the precise location of the users. The traditional global positioning system (GPS) uses satellite signals for the precise positioning of the user but is vulnerable to signal blockage in the complex indoor environment. The unforeseeable propagation losses due to multi-path effects as well as the permittivity and permeability difference of the materials lead to non-linear attenuation in the electromagnetic (EM) beam generated by the beacon devices in the indoor environment. Therefore, a robust indoor localization algorithm is required to precisely localize the users in the indoor environment with severe EM blockages. In this paper, we propose a novel hybrid RS-DeepNet framework that uses received signal strength (RSS) from WiFi devices for indoor localization of users. The proposed RS-DeepNet is a deep learning architecture that utilizes multiple gated recurrent layers (GRU) and a K-nearest neighbors (KNN) classifier to estimate the precise location of the user in the indoor setup. Simulation results show that the proposed RS-DeepNet outperforms the state-of-theart approaches and efficiently localizes the users in two indoor scenarios and achieves a lowest mean absolute error of 4.81 and 1.68 meters, respectively.