OpenEducation

Home Page
Roadmaps
Uploads
Quiz Example

Inspiration

As students, our dev team has all benefited from spending time with tutors, whether private tutors or student TAs. All students stand to benefit from individualized tutoring, but not all students have access to it. That's why we created OpenEducation!

What it does

OpenEducation lets users submit their class materials (slides, code snippets, etc.), generates a relevant learning plan to gauge a student's understanding and guide them, and then uses an ElevenLabs agent to tutor users to help them achieve mastery.

How we built it

We built OpenEducation with a Next.js frontend and Python backend. We used AMD's Developer Cloud tools to host a VLLM for inference. This VLLM generates quiz questions based on user-uploaded content and maintains JSON files that store the level of mastery users achieve on different topics for each class. We use ElevenLab's voice agent to discuss course material with students in their native language, much like a tutor would, in order to coach and engage students in conversation.

Challenges we ran into

Getting the ElevenLabs voice agent system prompts to work properly was not something our team had much experience with, and generating course content via vLLM prompting was cumbersome, and it was difficult to achieve consistent performance.

Accomplishments that we're proud of

Johan - This was my first hackathon with Next.js, and I was proud to get the ElevenLabs voice agent to integrate so seamlessly with the rest of the features Hudson - Designing the vLLM interface and establishing the link to our LLM, in addition to patching in responses to our JSON database for a more seamless experience. Raphael - Designing the system and getting the MVP up quickly, in addition to designing the frontend of the website.

What's next for OpenEducation

We want to deploy this website and raise funds for the hosting fees on a non-profit model, to increase the world's access to education. The best education is #OpenEducation