A machine learning–based SMS spam detection system designed specifically for Swahili language messages.
The system analyzes incoming SMS content in real time and classifies messages into three decision tiers: CLEAN, CONTENT_WARNING, or BLOCK, automatically applying clear and culturally relevant Swahili warning labels when needed.
It features a two-party SMS simulation interface that mimics real-world sender and receiver interactions, making it easy to test complete message flows and observe filtering behavior instantly.
Built with a FastAPI backend and a React-based frontend, the system achieves sub-150ms response times with over 86% classification accuracy, and includes a live dashboard for monitoring message statistics and system health.
The project demonstrates practical ML-powered content moderation for African language contexts, with a focus on clarity, speed, and real-world usability.