Biodegradable polymer composites have emerged as a sustainable alternative to petroleum-based materials in packaging, biomedical, and structural applications. However, traditional formulation techniques for reinforced polymer composites often lack precision and fail to adapt to real-time variations during processing, resulting in suboptimal material performance. This research proposes a real-time AI-IoT-enabled framework to optimize biopolymer composite formulations. The goal is to intelligently tune composite properties such as mechanical strength, moisture resistance, and biodegradation behavior by leveraging continuous sensor data and machine learning. A hybrid prediction model integrating Convolutional Neural Networks and Long Short-Term Memory (CNN-LSTM) with Extreme Gradient Boosting (XGBoost) was developed to predict composite behavior from IoT-based fabrication data. Key parameters polymer-filler ratio, temperature, humidity, and curing time were collected using an embedded sensor network during the composite processing stage. A Multi-Objective Genetic Algorithm (MOGA) was employed to optimize formulation targets across multiple property dimensions. Experimental validation was conducted using PLA-starch and PHA-lignin biopolymer composites. The proposed system achieved high predictive accuracy (R² > 0.90) across all polymer composite properties. Optimized reinforced formulations resulted in a 17.8% improvement in tensile strength and a 22.1% reduction in water absorption, while maintaining biodegradation above 90%. Experimental outcomes closely matched model predictions with less than 5% deviation. This study demonstrates a novel AI-driven optimization platform for biodegradable polymer composite formulations, offering a closed-loop, scalable solution for intelligent material design. The framework enables real-time formulation control for reinforced polymer systems, bridging performance, sustainability, and smart manufacturing.