This study presents the development and validation of a mobile-to-cloud road surface condition monitoring system using citizen-sourced smartphone sensor data in the Philippines. The proposed framework integrates mobile data acquisition, local preprocessing, secure cloud transmission, supervised classification, and geospatial visualization within a unified architecture. Tri-axial accelerometer and gyroscope signals were collected at 50 Hz, segmented into overlapping windows, and transformed into statistical feature vectors prior to cloud-based inference. Field deployment was conducted across 48 urban and peri-urban road segments, generating 12,485 processed feature windows. A labeled subset of 4,200 windows was used for supervised evaluation. The classification model achieved an overall accuracy of 88.4%, with balanced precision and recall across smooth, moderate, and severe surface condition categories. Confusion matrix analysis showed that misclassifications were primarily concentrated between adjacent condition levels rather than between extreme classes. System-level evaluation demonstrated near-real-time responsiveness, with an average end-to-end latency of 1.8 seconds under stable network conditions. Offline buffering achieved 100% synchronization success following connectivity restoration, ensuring data integrity in environments with intermittent network coverage. The results indicate that smartphone-based vibration sensing, when integrated with cloud analytics and geospatial visualization, provides a scalable and cost-efficient approach for preliminary road surface monitoring. The proposed framework offers a practical complement to conventional inspection methods and supports data-driven infrastructure prioritization in developing urban contexts.