StudyMate — An AI Study Partner Built for a Friend

Jaswanth Kumar built StudyMate, an AI study companion that extracts text from uploaded PDF notes, chunks it, and uses TF-IDF retrieval to feed relevant sections to an open-weight Qwen model for answers, explanations, and quizzes. The app pairs a React/TypeScript/Vite frontend with a FastAPI backend, stores document chunks in MongoDB Atlas, and runs Qwen3-4B-Instruct-2507 via Hugging Face Inference Providers with an option to run models locally through Ollama. It is deployed as separate backend and frontend services on Render.

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend https://dev.to/challenges/hacktoberfest-weekend-2026-10-01 I built StudyMate , an AI-powered study companion designed for a real friend who wanted a simpler way to understand class notes and prepare for exams. The problem was simple: students often have long PDF notes, but finding the right information and understanding difficult topics can take a lot of time. StudyMate lets them: The idea was not to build another general-purpose chatbot. I wanted to build something focused on one real person's study workflow: Notes → Understanding → Practice 🌐 Live Demo: https://studymate-local-frontend.onrender.com/ https://studymate-local-frontend.onrender.com/ The live version is deployed and can be used directly in the browser. 💻 GitHub Repository: https://github.com/Jaswanth-Kumar-2007/StudyMate-Local https://github.com/Jaswanth-Kumar-2007/StudyMate-Local The complete source code for the frontend and backend is available in the repository. StudyMate is built with React + TypeScript + Vite on the frontend and FastAPI + Python on the backend. The main AI pipeline is: PDF Study Notes ↓ PDF Text Extraction ↓ Text Chunking ↓ TF-IDF Retrieval ↓ Relevant Study Sections ↓ Qwen Open-Weight Model ↓ Answer / Explanation / Quiz the deployed version, I use Qwen/Qwen3-4B-Instruct-2507 through Hugging Face Inference Providers. I also designed the project so that the AI layer can be run locally using Ollama with an open-weight Qwen model. Tech Stack Frontend Backend AI Database Deployment MongoDB is used to persist the extracted document information and text chunks. Open innovation made it possible for me to build StudyMate around technologies that I can experiment with, understand, and adapt instead of depending entirely on a closed AI system. The project uses the open-weight Qwen model as its AI foundation. For the deployed application, Hugging Face provides convenient inference, while the same project can also be connected to Ollama for local model execution. This gives StudyMate flexibility to: The project also relies on open-source technologies such as React, FastAPI, pypdf, scikit-learn, PyMongo, Vite, and Lucide React. StudyMate is deployed using Render , with the FastAPI backend and React frontend hosted as separate Render services. Render made it possible to deploy the complete application and make the study companion accessible through a public web interface. StudyMate uses MongoDB Atlas as its data layer . When a student uploads a PDF, the application extracts the text, splits it into study-note chunks, and stores the document information and extracted chunks in MongoDB Atlas. These stored chunks are then used by the retrieval pipeline to find relevant study material when the student asks a question, requests an explanation, or generates a quiz. This is an individual submission by Jaswanth Kumar . DEV Profile: https://dev.to/jaswanthkumarkamireddi https://dev.to/jaswanthkumarkamireddi