TravelCheckinApp: A Full-Stack Smart Travel & POS Ecosystem

Mai Nhựt Minh — Thu, 28 May 2026 03:49:28 +0000

This is a submission for the GitHub Finish-Up-A-Thon Challenge.

What I Built

TravelCheckinApp ("SmartData AI-Flow") is an ambitious, production-scale travel management ecosystem designed for my graduation thesis at Tay Do University.

Unlike fragmented tourism applications, this platform seamlessly bridges the gap between tourists and local merchants by integrating a consumer-facing travel discovery app with a fully functional Point-of-Sale (POS) and Property Management System (PMS).

I. Codebase Metrics At a Glance

a. Total Source Code: 94,803 lines of pure TypeScript (Direct MySQL queries, zero ORM abstraction for maximum query optimization).
b. Architecture: Monorepo architecture controlling 3 sub-projects (Backend, Web, Mobile).
c. Backend API: 292 REST endpoints powered by Express 5 & Node.js.
d. Database: 50 tables handling complex business logic relationships.
e. Frontend UI: 67 reactive pages built using React 19, Vite 7, and Ant Design 6.

II. The Technical Stack

a. Frontend Website: React 19, Vite 7, TypeScript, Ant Design 6, TailwindCSS, Leaflet Maps.
b. Backend Infrastructure: Node.js, Express 5, TypeScript.
c. Database Layer: MySQL 8.0 (Utilizing advanced native spatial POINT types for SOS tracking, generated columns, and JSON arrays).
d. Real-time Engine: Hybrid architecture utilizing Socket.IO for bidirectional session management and Server-Sent Events (SSE) for lightweight booking state transitions.
e. Third-Party Integrations: VietQR (Dynamic QR code payment verification via bank transfer), Firebase Cloud Messaging (FCM) for push alerts, and Nodemailer.

III. Key Features Implemented

1. Consumer Features (For Tourists)

a. Interactive Discovery: Leaflet maps with custom filterable pins and real-time route polyline rendering.
b. VietQR Payments: Instant bank transfer processing using dynamic QR codes tracked live via SSE channels.
c. GPS-Verified QR Check-in: Location check-ins bound to physical geographic constraints (coordinates logged securely).
d. Gamified Exploration: Province-based check-in leaderboards, group sharing sessions (via 6-character shortcodes), and travel diaries with mood tracking.
e. Emergency SOS Chain: One-tap SOS signal blasting live coordinates to the admin center utilizing MySQL spatial indexes.

2. Merchant Features (For Business Owners)

a. Hotel PMS: A visual floor-plan grid system displaying real-time room status (vacant, occupied, reserved, cleaning) with inline check-in/out and stay extensions.
b. Restaurant POS: Interactive table maps with area filters. Waitstaff can handle dynamic menu ordering and billing at the counter.
c. Granular Staff Control: Owners can delegate scoped access keys to employees, routing them straight to operational front-offices while hiding sensitive revenue dashboards.

3. Administrative Features (For Platform Admins)

a. Operations: Comprehensive system analytics and centralized AI configuration toggles.
b. Moderation: Content moderation queues, verification review flags, and global platform voucher management.

Demo

Feel free to explore the production architecture, check out the raw SQL schemas, or contribute to the monorepo:

👉 GitHub Repository Link

The Comeback Story

This project originally kicked off in January as my core graduation thesis. However, due to an intensive university course schedule, development stalled, leaving behind separate layers of backend logic and fragmented administrative layouts.

The GitHub Finish-Up-A-Thon provided the exact spark I needed to "revive" the codebase. Within the timeline of this challenge, I pulled the frozen code out, established a strict monorepo layout, and achieved massive progress:

Core API Stabilization: Debugged and stabilized 80% of the entire Core Backend API and Web Admin interfaces (including cross-location POS transaction tracking and hotel reservation checkout constraints).
Background Worker Engine: Designed a robust core background worker cluster running intervals to auto-expire stale reservations every 60 seconds and dispatch push notification alerts.

My Experience with GitHub Copilot

Leveraging advanced AI assistance during this hackathon was a game-changer for rapid scaling:

Accelerated Refactoring: AI drastically cut down time when refactoring raw SQL prepared statements across huge controllers (adminController spans 11,600 lines; ownerController sits at 15,800 lines).
Architecture Blueprints: It helped design the cross-location verification protocol for our 3-layer signed HMAC-SHA256 QR token chain.
Error Remediation: Assisted in mapping out complex JSON arrays inside the MySQL table structures without breaking type safety.

IV. Future Vision & Roadmap

Phase 2 (Mobile App Deployment): Code the functional UI screens on top of my existing Expo SDK 54 mobile scaffold using expo-camera for QR verification and Zustand for state management.
Phase 3 (AI Microservice): Launch an isolated Python/FastAPI service dedicated to hosting 6 core modules, including a database-aware RAG Chatbot (using Google Gemini API) and automated collaborative filtering travel recommendation models.
Phase 4 (Production Hardening): Swap out current browser sessionStorage authentication tokens for secure httpOnly cookies, implement database connection pooling, and achieve comprehensive unit test coverage.

Thank you to DEV and GitHub for hosting this challenge and pushing me to bring SmartData AI-Flow back to life!

Let's connect! 👾 My GitHub Profile

DEV Community: Mai Nhựt Minh