AI-Powered Learning Platform

Full-stack LMS with context-aware AI tutoring and persistent RAG-lite capabilities.

🏗 System Architecture

• Backend: Java 17 / Spring Boot (REST API)
• Frontend: React (SPA)
• Database: PostgreSQL (Relational persistence)
• Security: Spring Security + BCrypt (Stateful/Session-based)
• AI Integration: Section-contextual prompt engineering via external LLM API

---

🛠 Technical Implementation

🔐 Security & Auth

• Authentication: Custom UserDetailsService implementation with BCrypt password encoding.
• Authorization: Role-based access control (RBAC) protecting course management and AI endpoints.
• Global Exception Handling: @ControllerAdvice mapping domain exceptions to standard HTTP status codes.

🧠 Context-Aware AI Logic

• The "Section-Aware" Engine: The backend fetches the current Section content (Markdown) and injects it into the LLM system prompt, ensuring responses are grounded in the specific lecture data.
• Persistence: Chat sessions are mapped to (User, Section) pairs in PostgreSQL, allowing for historical retrieval and multi-session management.

💾 Data Modeling

• Hierarchical Content: Course 1 -> N Sections.
• Chat Schema: ChatSession holds metadata; ChatMessage holds the content and role (User/Assistant) with timestamps for chronological rendering.
• Migrations: Transitioned from H2 (volatile) to PostgreSQL (persistent) for production readiness.

---

🔌 API Overview

Endpoint	Method	Description
/api/auth/**	POST	Login/Registration (Public)
/api/courses	GET	List all available modules
/api/sections/{id}	GET	Retrieve Markdown content for a section
/api/ai/chat	POST	Send message + Contextual metadata
/api/ai/history/{sectionId}	GET	Fetch persistent chat history

---

🚀 Deployment & CI/CD

• Pipeline: GitHub Actions automates mvn clean verify and frontend builds on every push.
• Environment: Production-ready configuration using Spring Profiles (application-prod.yml).
• Logging: SLF4J/Logback integration with severity filtering.

---

🛠 Setup

# Backend
./mvnw spring-boot:run -Dspring.profiles.active=dev

# Frontend
npm install && npm start

---

📈 Future Scope

• Vector Search: Moving from simple context injection to RAG using a Vector DB.
• Analytics: Tracking student engagement via AI query patterns.