This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend I built FriendStudy AI, an AI-powered study companion designed to help students study more effectively.
It provides students with an interactive study workspace where they can:
The idea came from a simple problem: students often have to switch between multiple tools for planning, learning, practicing, and tracking their progress.
FriendStudy AI brings these features together in one place.
I built FriendStudy AI for students who want a simple study companion that can help them understand topics, practice questions, and stay consistent with their studies.
The goal was to create something that feels less like a generic chatbot and more like a study companion that students can use as part of their daily routine.
Live Demo:
https://friendstudy-ai-hf26.vercel.app/ The AI features are powered by Qwen 2.5 3B, running through Ollama.
The application uses:
The production architecture looks like this:
React → Vercel → FastAPI → Render → ngrok → Ollama → Qwen 2.5 3B
The frontend communicates with the FastAPI backend, while the backend sends AI requests to the locally hosted Qwen model through Ollama.
Students can ask questions and get AI-generated explanations using Qwen 2.5 3B.
FriendStudy AI can help students organize their study sessions and create structured study plans.
Students can practice topics using AI-generated quiz questions.
The dashboard provides an overview of study progress, completed topics, daily goals, and study streaks.
Students can set study goals and keep track of their daily progress.
This project helped me learn a lot about taking an AI application from local development to a deployed application.
Some of the biggest things I learned were:
The biggest challenge was getting the local Qwen model to work with the deployed application.
Initially, the frontend and backend worked separately, but the deployed backend could not directly access Ollama running on my computer.
I eventually built this architecture:
Getting every part of this pipeline working together required quite a bit of debugging.
I also had to troubleshoot frontend build issues, CORS configuration, environment variables, API connectivity, and the connection between the deployed backend and the local AI model.
Seeing the deployed application successfully send a request through the entire pipeline and receive a response from Qwen made the debugging worth it.
The complete project is available on GitHub:
https://github.com/arkendukundu-dev/friendstudy-ai-hf26 Feel free to explore the code and try the application.
I would like to continue improving FriendStudy AI by adding:
Building FriendStudy AI was a great experience because I wasn't just building another AI chatbot.
I wanted to build something that could actually become a useful study companion for students.
This project also gave me hands-on experience with AI, React, FastAPI, deployment, open-weight models, Ollama, and connecting local AI infrastructure with a deployed web application.
Thanks for checking out FriendStudy AI!
If you try it, I'd love to hear your feedback.