How keerthi-reddy/techmate works
Project type and purpose
• From the file tree and README, this is a full-stack web application with a React/TypeScript frontend, a Spring Boot backend (currently scaffolded under /backend), and a separate Python-based LLM microservice under /llm-service.
• The frontend handles user interaction (login, dashboards, role selection), the Spring Boot service will handle business logic and data persistence, and the LLM service provides AI/LLM-powered features.
Top-level components (based on folder structure)
• Frontend (frontend/)
– React + TypeScript, built with Vite
– Uses Firebase for authentication (firebase.tsx)
– Pages: Grandkid_View, AuthPage, Dashboard, EZMode, TapTutor, etc.
• Backend (backend/)
– Spring Boot application (to be fleshed out)
– Exposes REST APIs for CRUD and orchestration
• LLM-service (llm-service/)
– Python Flask or FastAPI (app.py)
– Provides AI endpoints for chat, feedback saving, etc.
• Assets & docs
– docs/: design notes, frontend-creation guide
High-level interactions
1. User ↔ Frontend
• Browser loads static HTML/CSS/JS from frontend (Vite build)
• User signs in via Firebase Auth SDK; obtains JWT token
2. Frontend ↔ Backend
• Frontend attaches Firebase JWT to HTTP requests
• Backend Spring Boot validates token, processes requests, reads/writes DB
3. Backend ↔ LLM Service
• When AI-powered features are invoked, the Spring Boot service issues HTTP(s) calls to the LLM microservice (JSON over REST)
• LLM service returns generated text, which backend may post-process or store
4. Backend ↔ Database (implicit)
Architectural patterns & design principles
• Separation of concerns: UI vs business logic vs AI logic
• Microservices: LLM service is its own deployable unit
• API gateway style: Spring Boot acts as single backend API for the frontend
• Token-based authentication: Firebase JWT validation in backend
• Tech stack boundaries: React/Vite vs Java/Spring vs Python