Social engineering attacks such as phishing, pretexting, and baiting exploit human trust and pose a significant risk to people and organizations. Traditional forms of cybersecurity training are static, passive, and text-based, and are not adequate to prepare users for actual threats. With this in mind, this paper presents CyberShield, a gamified, AI-backed web-based training platform to make users more resistant to social engineering. CyberShield integrates interactive training, NLP-based personalization, reinforcement learning for adaptive situations, and 3D AR visualizations on a bleeding-edge tech stack with Streamlit, Three.js, and Plotly. The platform uses community-based intelligence and real-scenario simulations to provide situational, immersive, and inclusive training, conforming to WCAG 2.1 AA guidelines. Evaluation results indicate a 45% increase in user engagement, an 80% completion rate of training, a 60% increase in phishing and baiting threat detection, and a 25% increase in knowledge retention through gamification. This work adds to cybersecurity education by proposing a scalable, user-centered solution superior to traditional methods of training and promoting best practices in workplace and educational environments.