← All projects

Adaptive Online Learning Platform

BSc project · State University of Trade and Economics, Kyiv, 2024

A web platform that personalises each learner’s path using adaptive content and assessment driven by AI. The ML pipeline uses MBart for summarisation across languages, SentenceTransformer semantic similarity grading that beats exact match baselines on paraphrased answers and the OpenAI API for adaptive question generation, feedback and reading recommendations.

I started from an analysis of functional and nonfunctional requirements. The platform has secure authentication and role management for students, teachers and administrators. Learning management covers scheduling, assignment submission, automated testing and progress tracking, and personal profiles help users plan their work. Integration with sources such as YouTube and Wikipedia gives quick access to supplementary material.

For the nonfunctional requirements I added data encryption and optimised database queries and caching. A modular architecture makes new features easy to add. Error handling includes detailed logging and clear messages for users.

I modelled the system with UML: use case diagrams for student and teacher roles and sequence diagrams for workflows such as retrieving schedules and managing assignments. An adaptive learning algorithm analyses student responses, finds knowledge gaps and recommends next steps.

The system has three layers. The client uses HTML, CSS and JavaScript. A Django server handles business logic and security, and PostgreSQL stores the data.

The code is on GitHub.