A personal tool for summarizing large English PDFs with AI. SkimPDF breaks down your documents into manageable chunks and uses transformers to generate concise summaries — perfect for skimming research papers, reports, or any lengthy document.

• ✅ Upload any English PDF and get a summarized output
• ✅ Smart token-aware chunking to handle large files
• ✅ Abstractive summarization powered by facebook/bart-large-cnn
• ✅ Markdown-based parsing with pymupdf4llm for structured extraction
• ✅ React-based sleek frontend (in separate repo)
• ✅ CORS-enabled API for easy frontend integration

git clone https://github.com/Anand-Raut/skimpdf.git
cd skimpdf

pip install -r requirements.txt

uvicorn main:app --reload

Returns:

{ "isrunning": true }

Body: multipart/form-data with field pdfFile

Response:

{
  "filename": "your_uploaded_file.pdf",
  "status": "received",
  "summary": "Summarized text here..."
}

1. User uploads a PDF from the React frontend
2. Backend converts PDF to Markdown via pymupdf4llm
3. Text is chunked based on token count using tokenizer logic
4. Each chunk is summarized using facebook/bart-large-cnn
5. Summaries are merged and sent back as the final response

SkimPDF is designed for local use only. Run it on your machine, connect it with the frontend, and you’re good to go.

• Async background summarization
• Smarter semantic chunking
• PDF upload frontend (styled, responsive)
• Optional Docker support

Anand Raut GitHub Profile

MIT License — Open source, use it, tweak it, improve it. If it breaks, you get to keep both pieces.