<span><span>SOSENS: A Machine Learning-Enabled Soil Quality Monitoring and Decision Support System for Climate-Smart Agriculture in Sub-Saharan Africa</span></span>

Soil degradation poses a critical and escalating threat to agricultural productivity across Sub-Saharan Africa, where more than 70% of Rwanda’s population depends on smallholder farming for subsistence. Despite this urgency, affordable and accessible soil quality intelligence remains structurally unavailable to rural farmers, who are constrained by the prohibitive cost of laboratory testing, the contextual mismatch of existing precision agriculture platforms, and limited rural digital infrastructure. This paper presents SOSENS, an end-to-end Climate-Smart Agriculture Decision Support System that integrates IoT-based soil sensing, machine learning inference, and multi-channel communication to deliver real-time, actionable agronomic recommendations to smallholder farmers. A Random Forest classifier is trained on a multi-source dataset comprising soil macronutrients (N, P, K), pH, and seasonal meteorological variables to predict crop suitability across five categories. Systematic hyperparameter optimisation via stratified 5-fold cross-validation yields a classification accuracy of 66.21%, with misclassification patterns demonstrating agronomic coherence rather than random error. The system is deployed as a three-tier cloud-native application, comprising a FastAPI backend, a PostgreSQL persistence layer, and a React single-page application, augmented by a multi-channel notification service achieving a 94% delivery success rate via SMS and email. Production evaluation confirms 99.2% system uptime and a mean inference latency of 1.8 seconds. User acceptance testing with a cohort of twelve farmers validates interface usability and reveals a strong demand for recommendation explainability, motivating a SHAP-based transparency roadmap. SOSENS demonstrates that context-sensitive, low-cost ML engineering can bridge the gap between precision agriculture science and the operational realities of rural Sub-Saharan African farming.