Gas Metal Arc Welding (GMAW) remains a critical manufacturing process across aerospace, automotive and construction industries, yet traditional quality control methods rely on manual inspection and subjective assessments, leading to inefficiencies, increased rework costs and potential safety compromises. This research proposes an integrated IoT-enabled Cyber-Physical System (CPS) framework designed to enable intelligent, real-time defect detection and quality monitoring in GMAW processes by combining multi-modal sensor fusion with advanced machine learning algorithms. The framework integrates heterogeneous sensor data streams—including electrical arc signals (voltage and current), thermal imaging, acoustic emissions and torch position sensors—through a distributed edge-cloud computing architecture. Advanced deep learning models, specifically embedded system for image-based defect classification, Long Short-Term Memory (LSTM) networks for temporal pattern recognition in arc signals and ensemble methods (XGBoost optimized with Particle Swarm Optimization) for multi-sensor data fusion, are employed for real-time anomaly detection and quality classification. The proposed work involves: (1) design and development of a cost-effective IoT-based multi-sensor acquisition system with standardized data protocols; (2) implementation of a hybrid machine learning architecture capable of detecting critical defects such as porosity, lack of penetration and burn-through with enhanced accuracy and minimal latency; (3) development of a digital shadow system enabling predictive analytics for process parameter optimization and preventive maintenance; and (4) validation through experimental trials on industrial GMAW setups. Expected outcomes include achieving greater than 95% defect detection accuracy, reducing quality inspection time by 70%, enabling real-time process adaptation and providing a scalable framework adaptable to diverse welding environments and materials. This research bridges the gap between Industry 4.0 manufacturing demands and practical implementation challenges, delivering a comprehensive solution for autonomous, intelligent quality assurance in modern welding operations.