Inspiration
Students often spend a lot of time searching through lengthy study materials, trying to understand difficult concepts, and creating questions to test their knowledge. I wanted to build a platform that makes studying more interactive by allowing students to learn directly from their own documents.
This inspired me to build Smart Study Portal, an AI-powered learning platform designed to help students understand their study materials, practice concepts, and track their progress.
What It Does
Smart Study Portal turns uploaded study materials into an interactive learning experience. Students can:
- Upload study materials: Upload PDF, DOCX, PPTX, and TXT files.
- Ask questions: Ask questions about their uploaded content and receive contextual AI-generated answers with relevant sources.
- Generate quizzes: Create multiple-choice quizzes based on their study materials.
- Track topic mastery: View quiz performance and mastery levels to identify topics that may need more attention.
How I Built It
I built the project with a Python FastAPI backend and integrated Google's Gemini API for AI-powered responses and quiz generation.
The document-processing pipeline extracts readable text from uploaded files and divides it into smaller chunks. These chunks are embedded and stored in ChromaDB, enabling similarity-based retrieval when a student asks a question. This retrieval-augmented generation (RAG) approach helps the AI generate answers grounded in the uploaded study material rather than relying only on general knowledge.
SQLite stores structured application data, including document information, quizzes, attempts, answers, and topic mastery records. Quiz grading is performed against the stored correct answers, and mastery tracking combines previous mastery with the latest quiz score to reflect performance over time.
Challenges I Faced
One of the main challenges was connecting the different stages of the learning pipeline: document parsing, text chunking, vector storage, retrieval, and AI response generation. Each stage needed to work together so that uploaded content could be used effectively for both question answering and quiz generation.
Another challenge was preserving useful source information, such as page numbers, while processing documents into smaller chunks. I also needed to validate AI-generated quiz questions and options so that the generated output could be stored and graded reliably.
Building the mastery-tracking feature also required separating quiz evaluation from progress tracking, allowing quiz results to update a student's recorded mastery.
What I Learned
Building this project helped me understand how to integrate large language models into a practical application, implement a RAG pipeline, work with vector databases, and design backend APIs that connect multiple services.
I also learned that an AI-powered application involves more than calling an AI API. Document processing, data validation, persistence, error handling, and reliable evaluation are all important for creating a useful learning experience.
What's Next
I want to improve the platform with more advanced adaptive learning, including identifying recurring knowledge gaps, recommending targeted revision, and adjusting future learning activities based on student performance.
My goal is to make Smart Study Portal a practical study companion that helps students move beyond simply reading material toward understanding, practicing, and measuring their learning.
Log in or sign up for Devpost to join the conversation.