ParkSmart

Campus parking optimizer combining ML predictions with walking distance analysis.

Overview

Students face two major challenges when commuting to campus: not knowing which parking lots are closest to their classrooms, and not knowing if lots will be full when they arrive. ParkSmart solves both by combining intelligent routing with machine learning predictions.

Users upload their class schedule (via .ics file), and the app identifies which parking lots they're permitted to use, ranks them by walking distance to their classroom, predicts availability using historical patterns, and recommends the best parking options along with an optimal departure time.

Tech Stack

Layer	Technologies
Frontend	React 18, TypeScript, Vite, Tailwind CSS v4, Radix UI, MUI
Backend	Python 3.12, FastAPI, SQLAlchemy, asyncpg, Alembic
ML	Prophet (time-series forecasting)
Auth	Supabase Auth
Deployment	Vercel (frontend), Render (backend API + cron jobs)
Database	PostgreSQL (via Supabase)

Project Structure

ParkSmart/
├── frontend/          # React SPA (Vite + TypeScript)
├── backend/           # FastAPI REST API
├── data/              # Data files and datasets
├── docs/              # Project documentation
├── guidelines/        # Development guidelines
└── render.yaml        # Render deployment config

Getting Started

See the individual setup guides:

• Frontend README — React app setup and development
• Backend README — API server, database, and cron jobs

Deployment

• Frontend: Deployed on Vercel with SPA rewrites
• Backend: Deployed on Render as a web service with scheduled cron jobs for data collection and forecast generation (see render.yaml)

Design

Figma Design