How iamnguyenvu/airline-service-assistant works
Project type and purpose
• This is a full-stack, AI-powered “Airline Service Assistant” platform combining a Next.js frontend (React App Router) with a Spring Boot backend.
• Purpose: Provide flight search/advisory, policy Q&A and real-time analytics via an AI chat interface, integrating LLM tool-calling, vector search (RAG), scheduled ingestion, and caching.
File structure
• Top level:
• backend/ (Spring Boot microservice)
• frontend/ (Next.js 16 App Router)
• docker-compose.yml (local orchestration)
• docs/ (architecture, deployment, development guides)
• backend: layered into ai/, flight/, policy/, user/, config/ and uses Flyway migrations, Redis cache, and REST controllers.
• frontend: organized into app/ pages, components/, lib/api client, hooks (React), public assets.
• CI/CD: .github/workflows for backend, frontend, Docker publishing.
• Docker Compose brings up PostgreSQL (with pgvector), Redis, Ollama (local LLM) plus services.
README for additional insights
• Articulates a modular architecture: AI Orchestrator + domain-specific tools (Flight, Policy).
• Core technologies: Java 21 + Spring Boot, Spring AI, Next.js, Tailwind & shadcn/ui, pgvector, Redis, Gemini API/Ollama.
Elements
• Frontend (blue)
• Backend core (green)
• Data stores (orange)
• AI engines (purple)
• External services (gray)
• Annotations for ports and tech (e.g., “Next.js (3000)”, “Spring Boot (8080)”, “PostgreSQL+pgvector (5432)”).