klaushofenbitzer/invoice_processing repository overview

Invoices are “structured” only in the human sense. In practice, they arrive as PDFs with inconsistent layouts, nested tables, stamps, logos, and formatting quirks that make traditional automation brittle.

This app provides a local-first invoice extraction solution that turns unstructured invoice PDFs into clean, analyzable spreadsheet data—powered by a local vision-capable LLM using either a local (e.g. qwen3-vl-8B or similar) or remote LLM.

Project on Github: https://github.com/khofenbitzer/invoice_processing.git⁠