indy-safety-chat

"SafeIndy: Smart Chatbots for Safer Streets, Faster Alerts, and Empowered Communities."

Comment

🛡️ About the Project: SafeIndy

🌟 Inspiration

Our inspiration stemmed from a critical need: bridging the gap between public safety services and the people of Indianapolis using everyday technology. With a population of over 870,000 residents, timely access to emergency alerts, hazard reporting, and community resources can be the difference between danger and safety.

We asked ourselves: What if public safety services could live in every resident’s pocket? That question led to SafeIndy, a chatbot-based platform that empowers communities through instant access to essential safety tools.

🧠 What We Learned

  • Designing chatbot conversations for clarity, urgency, and empathy
  • Ensuring inclusive access through SMS, web, and WhatsApp interfaces
  • Integrating real-time APIs (e.g., NOAA for weather alerts, city services APIs)
  • Implementing LLM-based intent recognition with fallbacks for emergency scenarios
  • Prioritizing security, rate limiting, and privacy-first design for citizen data
  • The importance of multilingual and accessible design in public infrastructure

🏗️ How We Built It

  • Frontend: Built with React.js for the web interface and Twilio for SMS and WhatsApp integration.
  • Backend: Developed using Node.js and Express, with RESTful endpoints for chatbot interaction and alert broadcasting.
  • Chatbot Engine: Leveraged Rasa for intent classification and fallback handling; optionally integrated GPT-4 with safety guardrails for advanced Q&A.
  • Database: PostgreSQL for structured reporting and real-time status tracking.
  • Deployment: Hosted on AWS Lambda and Vercel for scalability, with HTTPS and JWT-based admin access.
  • APIs Used: NOAA Weather Alerts API, OpenStreetMap, City of Indianapolis 311 service, and more.

🚧 Challenges We Faced

  • Balancing speed with accuracy in emergency contexts — the bot had to be fast, but never mislead.
  • Integrating with local government systems, some of which lacked modern APIs.
  • Building fallback systems for offline and low-bandwidth access via SMS.
  • Ensuring data privacy and user trust, especially for anonymous crime tip features.
  • Multilingual NLP training to support diverse communities in Indianapolis.

🌆 Impact

SafeIndy aims to enhance civic trust and public safety by providing instant access to services, reducing emergency response lag, and empowering individuals with accurate, localized safety information — anytime, anywhere.

Built With

Share this project:

Updates