Recall Rehearsal

Inspiration

Recall Rehearsal is a study VR application aimed on the Feynman learning technique. A style of learning centered around being able to teach others to identify gaps in your own knowledge.

How we built it

Recall Rehearsal has 3 major parts. A flashy front-end site for uploading your notes to our backend for processing, a powerful personally-trained RAG model for context-aware feedback and question generation, and a full fledged VR app to immerse yourself while reviewing.

For our website, we utilised React Three Fiber, javascript, and react along with all custom made 3D models, scenes, and animations made in spline and blender to make our PDF processing process as fun as possible. Once you upload your PDF, it is sent to our backend that was built with Python and FastAPI. With langchain, we parse and tokenize the PDF and create a ChromaDB vectore store of embeddings to power a personalized RAG to optimize OpenAI's GPT-4o model. Once the PDF is done being processed, you are redirected to our VR app that we built with Unity for you to begin Recalling and Rehearsing!.

Tech Stack:

• React
• JS
• React Three Fiber
• Spline/Blender
• Python
• Fast API & ngrok
• LangChain
• OpenAI API
• ChromaDB
• VR app in Unity with C#

Challenges we ran into

• Never made a VR app before so many roadblocks
• Wifi issues and request issues with the oculus quest 2
• 3D Animations!

- First time implementing RAG

Accomplishments that we're proud of

We have never made a VR app before so we are very proud of our final product and are glad to have accomplished all of our goals for this project.

What we learned

We all learned A LOT about 3D Design and Modelling, along with creating VR Games using unity.

What's next for Recall Rehearsal

Built With

• blender3d
• c#
• javascript
• python
• react
• reactthreefiber
• spline3d
• unity