Inspiration
Parking Pilot was born from a common urban frustration: the endless search for parking. It transforms this stressful task into a seamless, intelligent experience using AI and modern web technologies. What started as a simple idea has grown into a system showcasing how AI, real-time communication, and progressive web apps can solve practical urban mobility challenges.
What it does
Parking Pilot is an AI-powered smart city companion. See all nearby parking spots instantly, compare fees, and reserve your ideal space ahead of time—even hands-free via voice. Beyond parking, discover local hotspots with AI-generated vibe hashtags and get real-time guidance when the best choice isn’t driving at all. City life becomes effortless, spontaneous, and fun.
How we built it
- Frontend: React 18 with MapLibre GL JS, Socket.io, and PWA capabilities.
- Backend: Node.js microservices, Express.js APIs, Firebase for auth, WebSockets for real-time updates.
- AI Service: Python FastAPI, LangGraph, multi-LLM support (OpenAI, Anthropic, local models), MCP integration.
- Cloud: AWS EC2 hosting for AI models.
- Key Decisions: Microservice architecture for scalability, real-time WebSocket updates, API-first design, progressive enhancement, zero vendor lock-in. ## Challenges we ran into
- Seamless Booking and Navigation: Ensuring users can find and reserve the right parking spot quickly was complex. We solved this with real-time state synchronization via WebSockets and AI-assisted recommendation logic that considers distance, fees, and user preferences.
- Dynamic Destination Guidance: Users’ plans often change, requiring flexible guidance. The solution was AI-generated vibe hashtags and real-time route suggestions that adapt dynamically to conditions.
- Cross-Platform Consistency: Delivering a consistent experience on desktop, mobile, and tablets required progressive web app techniques, responsive design, and offline fallbacks. ## Accomplishments that we're proud of
- AI Integration: Implemented one of the first MCP-based real-world AI booking assistants.
- Zero-Dependency Mapping: Built full-featured mapping with OpenStreetMap, eliminating costly API dependencies.
- Multi-LLM Architecture: Flexible AI system supporting local and cloud LLMs for robust performance.
- Seamless User Experience: Intuitive UI, fast search (<500ms), offline support, push notifications. ## What we learned
- Distributed system design, AI integration, real-time synchronization, and cross-platform PWA development.
- API-first, flexible, and well-documented architecture accelerates development and feature expansion. ## What's next for ParkingPilot - Smart City Parking Companion
- Short-term: Payment integration, predictive AI availability, mobile apps, enhanced security.
- Long-term: Dynamic pricing via ML, IoT sensor integration, multi-city expansion, carbon footprint tracking.
Built With
- amazon-web-services
- aws-bedrock
- python
- react-native

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