By supplying processing and storage resources at the edge, Mobile Edge Computing (MEC) in conjunction by Unmanned Aerial Vehicles (UAVs) has become a viable paradigm to improve the abilities of wireless networks. Distributing computing jobs to suitable resources, also known as task offloading, is a crucial step in maximizing the effectiveness of MEC systems. Owing to the constantly changing and distributed characteristics of the system, task offloading becoming more difficult in multi-UAV scenarios when numerous UAVs are used to support computing workloads. We provide a deep learning method for multi-UAV-assisted Mobile Edge Computing workload dumping. Convolutional neural networks (CNNs) and recurrent neural networks (RNNs) are two examples of deep learning approaches that we use to learn effective job offloading decisions according to numerous context characteristics, such as UAV position, networking circumstances, and computational capacity. We provide an extensive analysis of current deep learning-based offloading techniques and assess their effectiveness using experimentation and simulations. The outcomes show that, in multi-UAV MEC settings, our suggested deep learning methodology outperforms conventional techniques, resulting in improved resource efficiency and decreased latency. By optimizing task offloading choices in dynamic and resource-constrained contexts, the research helps Mobile Edge Computing systems make effective use of UAVs.