🚗 CatchARide
Have an extra seat on a long car ride? Offer it to a fellow student and save on some gas money!
CatchARide aimed to be a student-to-student rideshare platform where drivers with extra seats could match with passengers headed in a similar direction. Think of it as a smarter, more campus-oriented carpooling app. Despite our team's ambition and creative ideas, the project ultimately didn’t reach MVP due to time constraints and integration difficulties — but we learned a ton in the process.
🎯 Problem Statement
Students often travel between campuses, hometowns, or events, and many drive alone with extra seats. CatchARide's goal was to create a system that connects:
- Drivers, who post their route and open seats
- Riders, who enter their ideal pickup and destination points
Our matching algorithm would pair riders to the most optimal drivers, potentially splitting fuel costs and reducing environmental impact.
🧩 MVP Goals
- User Authentication
- Signup/Login
- OAuth2 using Google accounts
- ID Verification
- Groq + LLaVA for document + face recognition
- Backup: ONID verification with OTP
- Route System
- Drivers list routes
- Riders search and reserve seats
- Driver-Rider Matching
- Proximity-based matching
- Potential integration of optimization algorithms
- Map Interface
- Google Maps API to view and filter routes
🧪 Features Attempted
| Feature | Status | Stack |
|---|---|---|
| User Signup/Login | ⚠️ Not fully functional | React + Google OAuth2 |
| ID Verification | ✅ Basic integration in backend | Groq (LLaVA-vision) + DeepFace |
| Map Display | ✅ Basic integration in backend | Google Maps API |
| Route Filtering | ❌ Not implemented | Planned via React filters |
| Posting Routes (Driver) | ⚠️ Partial integration | PostgreSQL backend |
| Route Reservation (Rider) | ⚠️ Partial integration | PostgreSQL backend |
| Matching Algorithm | ✅ Basic integration in backend | Planned: Google Route Optimization API / Hungarian Matching |
🔧 Tech Stack
Frontend:
- React
- Bootstrap
- React Icons
- Google Maps API
- Google OAuth2
Backend:
- Flask
- PostgreSQL
Verification & Matching:
- Groq (LLaVA 11b) for ID document & face match
- DeepFace (Python) for face recognition
- Google Maps Route Optimization API
- Matching algorithm ideas:
💥 Challenges & Why It Didn’t Work
- We hate React: None of us like working on frontend, and so while the backend implementation was pretty solid, we didn't have any working frontend besides a landing page.
- Integration overload: We thought it was simple to implement, but it was not. Route optimization is an extremely hard problem.
- Lack of planning: Due to a lack of time, we spent significantly less time planning than previous.
- No time for matching logic: We researched strong algorithms but didn’t have time to implement even a basic version.
- Coordination: Multiple directions were explored in parallel which diluted focus.
- Google Maps API: Although set up, real-time integration with routes and user input was incomplete.
- Poor physiological state: Consistent challenges and lack of sleep slowly destroyed our confidence, energy, and determination. A $50 API charge from Google Maps API completely demoralized one of our teammates.
🤔 Lessons Learned
- Get a basic version working first (auth + map), then iterate.
- Plan data models and flows clearly before touching code, and perform better requirements analysis to break the project down into bitesize chunks.

