As an AI Research Intern at AIST, I focused on developing cutting-edge defenses for federated learning systems, including the creation and testing of FedGrad, which significantly enhanced the security and reliability of federated learning under adversarial conditions.