Green & Gold Guru

GGG

Inspiration

We’ve all been there: it’s late at night, you’re stuck on a complex problem, and office hours are over. We realized that students often feel isolated when studying difficult subjects. We wanted to bridge the gap between "struggling alone" and "having a professor by your side," while injecting the school spirit that makes USF special.

What it does

Green & Gold Guru is a multimodal AI teacher by Rocky. Unlike chatbots that just give you the answer, GGG listens to you, and teaches you step by step using voice and visuals. It can listen, speak, and analyze writing and images. GGG allows you to work on a problem in real time using a whiteboard, chat, and voice to help you solve your problem by not giving you the answer.

How we built it

We built it with Streamlit as the main platform. Open router in order to get all our api keys. Edge TTS for our text to speech. Streamlit's canavas for the whiteboard, Speech recognition for voice recognition. pyPDF2 for analyzing the PDF's. numpy to process the images from the whiteboard

Challenges we ran into

Challenges were implementing each problem step by step. We were often getting stuck due to errors like AI crashing, choosing the right model where we first started with GPT4o but ended up changing to gemini due to Gemini being more accurate. Another problem we had was how to integrate a whiteboard which we ended up using a lot of libraries in order to implement each feature.

Accomplishments that we're proud of

We’re proud of this project because it represents our first fully integrated full-stack build. We successfully combined whiteboard, text, and speech modalities with a tutor system infused with USF spirit, creating a effective and engaging learning experience for USF students.

What we learned

Our biggest learning was how to build a full stack project as we often just built backend printing to console for our school projects. It was a amazing experience learning to use all these libraries and making a full project.