International Journal of Scientific Research and Engineering Development

Cloud-based Identity and Access Management (IAM) systems have become vital for securing user authentication, authorization, and access control across distributed environments. However, traditional IAM frameworks primarily rely on static policies and rule-based monitoring, making them vulnerable to sophisticated cyber threats such as insider attacks, credential misuse, and advanced persistent threats. To address these challenges, this research proposes a real-time anomaly detection framework designed to enhance the security of cloud-based IAM systems. The framework integrates machine learning models, specifically autoencoders and isolation forests to analyze user behavior patterns, detect irregular access activities, and initiate adaptive mitigation responses. By continuously learning from evolving access trends, the system effectively identifies deviations from established norms while minimizing false-positive rates. Experimental results demonstrate that the proposed framework achieves improved accuracy and detection speed compared to conventional IAM solutions. The incorporation of real-time analytics enables proactive defense mechanisms that respond dynamically to emerging threats without disrupting legitimate user operations. This study underscores the importance of embedding intelligent anomaly detection into IAM infrastructures to strengthen identity assurance, ensure data integrity, and support zero-trust security architectures. The proposed approach offers a scalable, efficient, and adaptive model suitable for modern multi-cloud and hybrid environments.