Efficient irrigation management remains a critical challenge in modern agriculture, where excessive water usage leads to soil degradation and declining crop productivity. This work proposes an Edge-Enabled Smart Irrigation System (ESIS) that integrates a multi-sensor IoT network with real-time edge intelligence to optimize water distribution at plant-level granularity. The system employs soil moisture, temperature, humidity, pH, and light-intensity sensors distributed across farmland, achieving an average sensing accuracy of 96.4%. Data are processed locally using an edge-AI module powered by a lightweight Gradient Boosting model, enabling ultra-low latency decision-making (<420 ms) without reliance on cloud connectivity. A dynamic irrigation control algorithm adjusts valve operations based on crop-specific thresholds, reducing unnecessary watering cycles by 38.7% when compared to traditional time-based irrigation. In field trials conducted over a 45- day crop cycle, ESIS achieved a 41.2% reduction in water consumption, a 27.6% improvement in soil moisture stability, and a 19.4% increase in overall crop yield. Network reliability was maintained at 98.1% through an adaptive mesh communication protocol using LoRa and Wi-Fi hybrid links. Additionally, the energy-efficient sensor design extended node lifetime to 142 days on a standard 2200 mAh Li-ion battery. The results demonstrate that the proposed ESIS framework provides a robust, scalable, and sustainable solution for precision agriculture, offering significant benefits for water-limited farming environments.