Global food demand continues to rise, requiring agricultural systems that conserve limited water resources while maintaining productivity. This paper presents an IoT-based smart irrigation framework adapted for crops grown in Vermicompost, a substrate with distinct water retention properties compared to conventional soil. Environmental parameters including temperature, humidity, and substrate moisture are measured using field sensors and transmitted to a Raspberry Pi gateway for processing and cloud storage. A decision tree classifier, trained on Vermicompost-specific datasets, predicts irrigation requirements with improved accuracy. Farmers receive real-time email alerts, and all sensor data is archived for analytics. Experimental results demonstrate enhanced irrigation scheduling, reduced water consumption, and improved crop health when Vermicompost-specific thresholds are applied. The proposed system highlights the potential of integrating machine learning with eco-friendly substrates to achieve sustainable and climate-resilient farming.