05-document-processing

Module 05 · Document Processing

Automate enterprise document intake: extract, classify, and structure data from PDFs, Word docs, and scanned images.

Files

File	Description
document_pipeline.py	Full pipeline: extract → classify → structure → PII detect

Quick Start

python 05-document-processing/document_pipeline.py

Pipeline Stages

1. Text Extraction — PDF (pypdf), DOCX (python-docx), Images (Tesseract OCR)
2. Classification — Zero-shot document type detection (HuggingFace)
3. Structured Extraction — LLM-powered field extraction (Grok/OpenRouter)
4. PII Detection — NER-based entity detection before storage

Supported Document Types

• Invoices → vendor, line items, totals
• Contracts → parties, dates, obligations
• Policies → requirements, owners, review dates

Enterprise Use Cases

• Accounts payable automation (invoice processing)
• Contract lifecycle management
• Compliance document review
• HR document digitization