Precision agriculture has emerged as a promising solution to address the challenges of declining soil health, inefficient resource utilization, and improper crop selection in traditional farming practices. Conventional soil analysis methods are often time-consuming, costly, and incapable of providing real-time insights required for timely decision-making. To overcome these limitations, this paper presents an AI-driven smart soil and crop recommendation system built on an embedded Internet of Things (IoT) platform.The proposed system integrates a multi-parameter soil sensor with an ESP32 microcontroller to continuously monitor critical soil and environmental attributes such as nitrogen, phosphorus, potassium, pH, moisture, and temperature. The sensed data is transmitted to a cloud platform using wireless communication protocols, where advanced machine learning models perform intelligent analysis. Transformer-based and deep learning models are employed to recommend suitable crops, optimal fertilizer usage, and irrigation schedules based on real-time and historical data. The generated recommendations are delivered to farmers through a user-friendly mobile application and IoT dashboard, enabling informed and timely agricultural decisions. Experimental evaluation demonstrates improved prediction accuracy, efficient resource utilization, and enhanced accessibility compared to conventional farming approaches. The proposed system offers a scalable, cost-effective, and sustainable solution for smart agriculture applications.