In February 2025, Ahmed Elgallad developed a new Critic network architecture for the WE-Autopilot/Red-Team repository, targeting the f1tenth_gym reinforcement learning environment. He designed a CNN-based state processing pipeline combined with fully connected layers to improve state-action value estimation, supporting more stable policy updates in autonomous driving simulations. Ahmed implemented the solution using Python and PyTorch, integrating the Adam optimizer with configurable learning rates and robust checkpointing to enhance training reproducibility and convergence. His work demonstrated depth in deep learning and reinforcement learning, addressing the need for reliable policy evaluation and enabling faster, more controlled experimentation without reported bugs.