01 / The problem
Studying from a PDF often means jumping between a document, notes, and separate revision tools. Scanned documents add another obstacle: there may be no searchable text at all.
02 / The approach
PDF Buddy brings document questions, quizzes, flashcards, and notes into one workspace. According to the repository README, answers use retrieved passages and identify their source documents. Multiple PDFs can be searched together.
03 / Inside the build
- A React and Vite frontend renders chat, Markdown, and mathematical notation.
- FastAPI manages document processing and the retrieval workflow.
- Tesseract and Poppler extract text from scanned pages.
- Sentence-transformer embeddings and FAISS provide document retrieval.
- The Groq API generates responses from the retrieved context.
- Notes use browser storage, including IndexedDB for images.
04 / Contributions and outcome
This project is published in Rudransh's GitHub account, with a linked hosted application. The README documents the application architecture and study features. A breakdown of personal contributions, user counts, and performance measurements has not been supplied.
05 / Lessons to document
The next update to this case study can describe retrieval quality, scanned-document edge cases, and trade-offs in the notebook's browser storage. These are areas exposed by the architecture, rather than claimed experimental results.