This repository implements a DDPG-based algorithm to minimize processing delay in UAV-assisted mobile edge computing systems. The system jointly optimizes user scheduling, task offloading ratios, and UAV flight parameters (angle and speed) in a continuous action space. The approach addresses non-convex optimization with high-dimensional state spaces using deep deterministic policy gradient reinforcement learning. Experiments compare against DQN baselines and demonstrate convergence to optimal offloading policies.