SafeCity — Production-Grade Multimodal Safety System
SafeCity began with a question I kept coming back to: can a phone notice signs of distress before someone is able to ask for help? With StackOverHack, I lead its development as a production-grade system that brings sound, motion, visual cues, and context into one tiered-alert workflow.
View GitHub repositoryWhat we built
The system combines audio, visual, motion, and contextual signals to identify possible distress and support tiered alerts.
It is engineered for real-time multimodal processing, backend orchestration, and scalable alert escalation.
At TIH–IIT Mandi, the project appeared under the team name Team City (Safe Stack Over Hack) and finished third.
The ₹25,000 award was shared by the team.
I lead the SafeCity team listed in the SmallAI programme and help shape the system's architecture, backend orchestration, and development.
Evidence and links
The organiser and certificates use a few versions of our team name. The organiser confirms the team placement; my IIT Mandi lead role comes from my own account, while the SmallAI programme separately names me as SafeCity's team leader.