Inspiration
As immigrants and children of immigrants, we know firsthand how overwhelming arriving in a new country feels. The information exists — but it's scattered across 30 government websites, Facebook groups, and word of mouth. We wanted to collapse that chaos into one conversation.
What it does
PassportAI lets any immigrant describe their background in plain language. In seconds, our AI builds their personalized Life Passport — a complete roadmap covering visa steps, matched scholarships, local community organizations, and a career path. We also built a TikTok-style Discover feed where users scroll through surprising "Did you know?" facts about immigrating to countries they may never have considered.
How we built it
- React + Vite frontend
- Gemini AI API for profile extraction and passport generation
- Static JSON database of scholarships, community orgs, and visa steps
- CSS animations for the interactive accordion passport and snap-scroll Discover feed
- Built entirely in 3.5 hours at the McGill Data Network Hackathon
Challenges we ran into
Prompt engineering the AI to return clean, structured JSON consistently across wildly different user inputs. Also fitting a full product — two distinct features, polished UI, and a working AI pipeline — into 3.5 hours.
Accomplishments we're proud of
The Discover feed was born as an idea mid-hackathon and we shipped it anyway. The passport interactive accordion genuinely feels like a product, not a prototype.
What we learned
How to split a build between two people with zero overlap and merge cleanly. How to prompt AI models for reliable structured output under time pressure.
What's next
Document upload, multilingual support, and expanding beyond Canada to France, Germany, and the UAE.
Built With
- dotenv
- express.js
- flagcdn.com
- google-gemini-ai
- javascript
- local-json
- localstorage
- lucide-react
- node.js
- react
- tailwind-css
- vite
Log in or sign up for Devpost to join the conversation.