How RamaLearn Works
The Platform Stack
RamaLearn is built on a modern, lightweight technology stack designed for reliability, security, and scalability. The platform uses a Python and Flask backend, which provides a fast and flexible foundation for handling user requests, managing sessions, and communicating with the AI layer. The frontend is built with standard HTML and JavaScript, keeping the interface accessible and responsive across different devices and screen sizes. Data is stored using SQLite for the current development phase, with a planned migration to a more robust cloud database solution as the platform scales.
All user passwords are hashed using bcrypt to ensure that sensitive credentials are never stored in plain text. Email verification is handled through the Resend API, and user authentication is managed via JSON Web Tokens with role-based route protection that separates student and teacher access levels within the application. Security and data privacy are treated as foundational priorities throughout the entire development process, not as features added on later.
The AI Layer
The AI tutoring layer is powered by a large language model accessed through an external API, which allows RamaLearn to generate intelligent, context-aware responses to student questions in real time. The system is designed so that the AI has access to relevant context from the current lesson plan or assignment when generating its responses, ensuring that the tutoring stays aligned with what the teacher is actually teaching. This is a key differentiator from generic AI chatbots — the responses are grounded in what the student is actually working on.
Cross-session memory is a planned feature that will allow the platform to track patterns in how individual students interact with the AI over time, giving both the student and their teacher a clearer picture of learning progress. The platform is currently hosted on a home server running Windows Server 2025 with Hyper-V virtualization during active development, with deployment to Amazon Web Services planned as the product approaches a public release.
Python / Flask
Lightweight, fast, and flexible. The backend handles routing, sessions, and all communication with the AI API.
Auth & Security
JWT authentication, bcrypt password hashing, role-based route protection, and email verification via Resend.
Cloud Infrastructure
Currently self-hosted on Windows Server 2025. AWS deployment is planned as the platform scales to production.
LLM Integration
A large language model API powers the tutoring layer, with context injection from teacher-uploaded assignments.