Real-Time Safety Buddy (for Students & Women)

Project Title: Guardian AI – The Proactive Safety Companion

1. The Problem: The "Panic Button" Paradox Most personal safety apps are reactive. They rely on a user being able to reach their phone, unlock it, and press a button during a moment of extreme stress or danger. For students and women commuting late at night, this manual step is often impossible or escalates the threat.

2. The Solution: Proactive "Guardian" Intelligence Guardian AI shifts the paradigm from reaction to pre-emption. Using the Google Gemini API, it acts as a virtual escort that understands context. It doesn't just wait for an SOS; it monitors the "vibe" of the journey through multimodal inputs (voice, text, and environmental images) and intervenes early.

3. Key Features Multimodal Threat Detection: Uses Gemini 1.5 Flash to analyze audio snippets for aggression or "shaky" speech patterns and images for environmental risks (poor lighting, deserted areas).

Three-Tier Escalation Protocol: * Level 1 (Safe): Passive monitoring.

Level 2 (Uneasy): Active suggestions (e.g., "Turn left toward the pharmacy, it's better lit").

Level 3 (Danger): Automatic SOS triggers and "Distraction Dialogue" (simulating a loud phone call from a family member to deter attackers).

Smart Safe-Words: Users can trigger an emergency silently using natural phrases like "The weather is strange today," ensuring they don't alert a potential threat.

MongoDB Integration: Stores crowdsourced "Safety Ratings" of local Pune routes, allowing the AI to warn users about high-risk zones in real-time.

1. Technical Stack AI Engine: Google Gemini API (Multimodal 1.5 Flash).

Database: MongoDB Atlas (for storing emergency contacts and route safety logs).

Frontend: [React / HTML5 / Flutter] with real-time status indicators.

Communication: Twilio API (for automated SMS/Call alerts to trusted contacts).

1. Impact Guardian AI provides peace of mind for Pune’s massive student and IT population. By automating the decision-making process during high-stress moments, it ensures that help is already on the way before the user even has to ask.

Built With

• typescript