StuTutor-AI

Landing Page
What StuTutor-AI can do
PDF file submission
Real-live chat
More Features!

Inspiration

As a student juggling multiple courses, we often found ourselves overwhelmed by dense textbooks and lengthy lecture notes. I'd spend hours searching for key concepts, trying to connect ideas across pages, and wishing I had someone to just explain it simply. That's when the idea hit me: What if I could upload a PDF and talk to an AI that understands it?

StuTutor-AI was born from a real pain point - the gap between static learning materials and dynamic understandings. We wanted to build a tool that doesn't answer questions, but teaches - like a patient tutor who's read your entire syllabus.

What it does

StuTutor-AI is an intelligent, AI-powered tutoring platform that transforms any PDF (textbooks, lecture notes, assignments) into an interactive study companion.

Key Features:

Upload PDFs and have them instantly analyzed
Ask questions in natural languages - get context-aware answers
Interactive chat with full conversation history
Organize study sessions by document
Real-time explanations, summaries, and insights powered by OpenAI

It's like having a 24/7 tutor who's already studied your materials.

How we built it

We built StuTutor-AI as a full-stack React application with a focus on speed, simplicity, and developer experience.

Tech Stack:

Frontend: React, TypeScript, Vite, Tailwind CSS, shadcn/ui, React Router
Backend: Node.js, Express, Supabase (Auth + Postgres), Multer
AI: OpenAI API (gpt-3.5-turbo), pdf-parse
Dev Tools: ESLint, Nodemon, Sonner (toasts), Lucide Icons

Challenges we ran into

OpenAI token limits - We consumed all of our tokens while developing the project.
NeuralSeek integration - Tried alternative AI provider but we were unable to send out the responses from NeuralSeek into the user's chat.
Silent PDF extraction failures - Sometimes the PDF may contain tough characters to extract.
Chat losing context on reload - We have all the chat logs stored in the database, but on refresh in the app it would not load them again.
Auth0 integration - We ran into errors, so we had to scrap this in the end.

Accomplishments that we're proud of

First AI answer from *my own uploaded PDF* — pure magic
Built a complete auth + database system with Supabase in under 2 hours
Smooth, real-time chat with streaming responses and Tailwind animations
Clean, responsive UI using shadcn/ui — looks professional out of the box
Fully persistent conversations — survive page reloads and server restarts

What we learned

Token limits hit fast in real AI apps - we learned the hard way that it would sometimes not load the entire PDF + chat history can vanish if not managed correctly.
Not every AI API plays nice under pressure - we tried integrating NeuralSeek for enhanced reasoning, but we weren't able to send it's responses back into the chat.
PDFs are wild - some are images, some are locked, some have weird formatting.
Hackathon time = ruthless prioritization - we had to manage software development under intense time constraints.