With the rapid development of the e-commerce industry, how to effectively allocate human resources to cope with the dynamic changes and complex demands of orders has become an urgent problem that e-commerce companies need to solve. To this end, this paper designs an e-commerce human resource allocation optimization model that integrates deep learning algorithms and combines it with a computer-aided decision system (CADS) for real-time decision support. First, the Long Short-Term Memory (LSTM) network is used to extract features and predict demand for time-series information such as employee work performance. Then, the Deep Q-Network (DQN) is combined to make intelligent decisions, and the resource allocation strategy is optimized through reinforcement learning to maximize the utilization of human resources and customer satisfaction. Finally, the constructed computer-aided decision-making system can monitor in real time and dynamically adjust the configuration plan based on feedback information. In the experimental conclusion, compared with traditional methods, the optimization model based on LSTM and DQN significantly improves the order prediction accuracy, increases resource allocation efficiency by about 16%, shortens customer waiting time by 40%, and reduces operating costs, verifying the efficiency and reliability of the model in practical applications. This model provides a new resource scheduling solution for e-commerce platforms, which has strong theoretical value and practical application significance.