Effective management of water resources has become increasingly important in contemporary agriculture as a result of growing water shortages and changing climatic conditions. This paper presents an AI driven intelligent smart irrigation system that integrates multi-sensor fusion with embedded machine learning to enable automated and adaptive irrigation control. The proposed system continuously monitors soil moisture level, ambient temperature, relative humidity, and light intensity using distributed sensing units deployed in the agricultural field. The acquired data from multiple sensors are integrated and pre-processed to form a unified representation that improves reliability and reduces the effect of noisy measurements. An embedded machine learning model is implemented on a Raspberry Pi platform to analyze the fused sensor data and predict irrigation requirements in real time. Based on the model’s output, the system automatically controls a relay-driven water pump, supplying water only when required. This approach minimizes manual intervention, prevents over-irrigation, and optimizes water usage while maintaining suitable soil conditions for crop growth. Experimental evaluation using representative sensor data demonstrates that the proposed system can effectively adapt irrigation decisions to changing environmental conditions. The results indicate improved water efficiency and operational flexibility, making the system suitable for sustainable and precision agriculture applications