All projects

2026 / Applied AI / Product

NoteWorthy

A system for converting handwritten notes into structured, styled PDFs and LaTeX documents.

A typeset document generated from source notes by NoteWorthy.
Generated PDF output from the live conversion pipeline.
Role
Full-stack developer
Stack
Next.js / TypeScript / Google AI Studio / OAuth / Docker / Cloud Run

Problem

Handwritten academic notes are difficult to search, restyle, and reuse. NoteWorthy converts source notes into structured documents without requiring the user to rebuild formatting manually.

NoteWorthy wordmark.

Constraints

  • Preserve mathematical notation and document hierarchy.
  • Compile untrusted document input inside an isolated environment.
  • Support sign-in and a web-first upload-to-export workflow.

Architecture

A Next.js application manages uploads, authentication, and document state. Google AI Studio extracts structure and notation, while a Dockerized LaTeX compiler on Cloud Run generates the final output.

Implementation

  • Integrated GitHub and Google OAuth.
  • Mapped model output into a controlled LaTeX document shape.
  • Containerized compilation and deployed the product across Vercel and Cloud Run.

Current Status

The application is deployed and supports note conversion into styled PDF and LaTeX output. Current work is centered on extraction consistency for dense notation and varied handwriting.

Next Steps

  • Improve structured extraction evaluation.
  • Add better correction workflows before compilation.
  • Explore semantic organization and retrieval across converted notes.

System

  1. Note upload
  2. OCR / vision model
  3. Structured document
  4. LaTeX compiler
  5. PDF export