Rubber

Inspiration:

Frustrated by the tedium of using AI to transcribe and assist with proof assignments, we built Rubber to ease the process.

What it does:

Rubber automatically converts handwritten assignment notes into LaTeX and uses assignment context to better aid and give feedback to the student in question with further support to live edit LaTeX in order to better showcase it's suggestions.

How we built it

We build it utilizing langchain -> and gemini for llm querying and image ocr from digital notes to LaTeX Then we utilized excalidraw to provide the user with a whiteboard, to edit and produce notes for proof solving. All packaged around a nestjs and tailwind styled app

Challenges we ran into

We ran into quite a few issues with managing to get extractions from assignment pdfs and getting to gemini to understand chronological order of elements being inserted into excalidraw canvas.

Accomplishments that we're proud of

The app itself looks and feels quite elegant and runs well even using gemini's latest free model. Furthermore, the use of excalidraw makes writing and editing smooth as butter.

What we learned

For some of us this was our first hackathon, we learned how to build and scale under strict deadlines and work alongside, LLM's in that process and rush to build.