Multiple-input multiple-output (MIMO) orthogonal frequency-division multiplexing (OFDM) is a cornerstone technology for achieving high data rates in modern wireless systems, such as 5G and Wi-Fi. However, realizing its full potential requires accurate channel estimation and signal detection, which are challenging tasks for conventional receivers, especially in complex environments. This paper proposes a novel deep learning (DL) based approach that treats the MIMO-OFDM receiver as an end-to-end black box. Our proposed deep neural network (DNN) directly maps the received signals from all antennas to the original transmitted data streams, implicitly performing both channel estimation and spatial demultiplexing. Simulation results demonstrate that our model achieves performance comparable to the classical minimum mean-square error (MMSE) detector, without requiring any prior knowledge of channel statistics. The proposed approach highlights the potential of deep learning to simplify receiver design and enhance performance in future wireless communication systems.