TWINIX
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

# MetaGuardRail – AI-Powered Digital Safety and Content Protection System **MetaGuardRail** is an intelligent Android-based digital safety application designed to identify, analyze, and protect users from potentially harmful digital content in real time. It combines on-device machine learning, OCR, text analysis, and AI-powered content understanding to detect unsafe material and provide context-aware protection while prioritizing user privacy. The application analyzes text, images, and screen content to identify potential risks such as explicit content, violence, harmful language, and suspicious links. It uses severity-based classification to categorize content into SAFE, REVIEW, WARNING, and HIGH_RISK levels, helping users make informed decisions. MetaGuardRail integrates technologies such as **Google ML Kit, Tesseract OCR, TensorFlow Lite, Vosk Speech Recognition, and Meta AI models** to support intelligent content analysis, including offline detection capabilities. Its Live Protection feature uses Android MediaProjection and foreground services to monitor screen content and display protective overlays when potentially harmful material is detected. ### Key Features - AI-powered harmful content detection and classification. - Context-aware analysis to reduce false positives. - Text and image safety analysis. - OCR-based extraction and analysis of on-screen text. - Violence and explicit-content detection. - Suspicious URL and link safety analysis. - Offline-first detection with optional cloud AI assistance. - Severity-based risk classification. - Real-time protective overlay with temporary snooze functionality. - English and Telugu speech recognition support. - Privacy-conscious processing with user-controlled cloud image analysis. ### Technology Stack - **Platform:** Android - **Languages:** Kotlin, Java, Python - **Machine Learning:** TensorFlow Lite, Google ML Kit - **OCR:** Tesseract OCR - **Speech Recognition:** Vosk - **AI Integration:** Meta AI models through a backend proxy - **Data Storage:** Android DataStore - **Architecture:** Android foreground services, MediaProjection and local analysis components ### Project Objective The primary objective of MetaGuardRail is to build a practical, intelligent and privacy-conscious digital safety solution that helps users identify potentially harmful online content, understand its context, and maintain a safer digital environment.