Published Paper Details:

Modern distributed systems demand dynamic and intelligent load balancing to ensure optimal performance, high availability, and fault tolerance. Traditional load balancers often follow static routing strategies that fail to adapt to real-time backend conditions such as latency spikes, load variation, or system failure. This paper proposes an AI/ML-based intelligent load balancer that leverages machine learning algorithms to select the most suitable backend server based on live metrics including load, latency, and failure rate. The proposed system integrates Decision Tree, Random Forest, and XGBoost models to predict optimal routing paths, supported by real-time health checks and CSV-based metric logging. Among the models tested, XGBoost demonstrated the highest accuracy and was deployed in a Flask-based microservice architecture for intelligent request redirection. The results indicate a marked improvement in success rate and resource efficiency when compared to traditional random backend selection. This work enhances routing precision, improves infrastructure reliability, and lays the foundation for intelligent backend management in cloud-native and large-scale distributed systems.