LearnMate — an e-learning platform

LearnMate was my BCA 8th-semester final project: a study platform that turns course material into summaries, flashcards, and quizzes, then schedules reviews with a spaced-repetition algorithm.

What it does

• Materials — create study material from existing sources and organise it by subject.
• Summaries — abstractive summarization through a dedicated AI microservice.
• Practice — flashcards and generated quizzes for active recall.
• Spaced repetition — an SM-2 based scheduler adapts review intervals to each learner.
• Progress — track study activity over time.
• Reports — generate DOCX reports from study data (chunked on the server).

Architecture

LearnMate is split into a Next.js client, a dashboard, a Fastify API, and a separate Python service for AI work. The Fastify backend owns the domain logic (auth, scheduling, DOCX generation, report chunking) over PostgreSQL via Prisma, while the Python service handles text cleaning, tokenisation, and summarization and returns results to the TypeScript API.

Layer	Choice
Frontend	Next.js, React, TypeScript, Tailwind, ShadCN
Backend	Fastify, Prisma ORM
Database	PostgreSQL
AI service	Python, FastAPI, Hugging Face Transformers (BART)
Auth	JWT

Repositories

Four repos: the LearnMate client, LearnMate-Dashboard, the learnmate-api Fastify backend, and learnmate-fastapi-microservice for BART summarization.

Known limitations

There is no live deployment — the repositories are archived. The report’s design is complete, but no follow-up work was done after the external evaluation.