server-side shows the app is running
store the data and send to consent party
locations where the person ,who wants help

What it is:

SafeWay is a safety app designed for when you're in an unfamiliar or risky area — and maybe don't have internet. It helps you send alerts, share your location, and find safer routes even without a network connection.

Why we made it

Lots of safety apps stop working when you lose internet. That leaves people vulnerable — travelers, solo commuters, elderly, or anyone in low-signal areas. SafeWay was built to solve exactly that: safety shouldn't depend on being online.

What it actually does for you

📡 Offline mode:

-Send SOS alerts via Bluetooth/Wi-Fi Direct to nearby SafeWay users

-Get basic safe-route guidance using just GPS and offline maps

-Works without internet

🌐 Online mode:

-SOS alerts forwarded in real-time to emergency contacts and cloud

-Full navigation with safety-rated routes

-3D/4D maps with safety insights

🤖 Smart safety checks:

Uses on-device AI to check routes for safety (crime rates, lighting, busyness) — no data leaves your phone, so it's private.

How we built it

-App: React Native (works on iOS and Android)

-Backend: Node.js for SOS relay when online

-AI: Trained using public datasets on IBM’s platform, then optimized to run locally on phones

-Maps & data: OpenStreetMap + public safety datasets

-Connectivity stack: Bluetooth LE, Wi-Fi Direct, GPS, and GSM fallback

Hard parts we faced

-Making Bluetooth work reliably for offline alerts (tiny data limits, phone compatibility issues) -Shrinking the AI model to run fast on phones without draining the battery -Handling all the phone permission prompts and different OS rules (Android vs. iOS) -What we learned -Building apps that work offline-first is tougher but way more useful -Optimizing AI to run on-device is key for speed and privacy -Real-world safety apps need careful balancing of features, battery life, and permissions

The math behind safer routes

-We calculate a safety score for every route using things like: -Crime reports in the area -Street lighting info

How busy the street usually is

-The score is a weighted formula — important factors (like crime density) affect the score more.

What’s next for SafeWay

-Connect with official emergency response systems -Support for smartwatches/wearables -AR navigation mode (“safety arrows” in real-world view) -Expand for global travelers and remote workers

Built With

client-side
computer-netwoks
ibm-cloud
java
javascript
mongodb
node.js
python
react-native
server-side

Submitted to

Gemini 3 Hackathon

Created by

I came up with the idea for SafeWay and built it from start to finish as a solo project.

My work included:

Identifying the problem — noticing that most safety apps fail without internet, and deciding to create an offline-first solution.

Designing the whole system — figuring out how the app would work offline and online, and how to keep it private and secure.

Building the mobile app — writing the code for both iOS and Android using React Native.

Developing the backend — setting up the SOS alert system using Node.js so messages could be relayed when online.

Implementing offline features — using Bluetooth, local storage, and Wi-Fi Direct to let people send alerts without internet.

Working with data & AI — picking safety-related datasets, designing the on-device AI model to rate route safety, and keeping everything private (no data sent to the cloud).

Handling the full scope — planning the architecture, creating diagrams, writing documentation, testing everything, and putting together demos for presentation.

shivacharan dhoni

Updates

shivacharan dhoni started this project — Feb 03, 2026 05:44 AM EST

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.