Project Overview
Automated Python verification pipeline that scanned and verified 600+ 30-page legal PDF agreements at SkipCash, compressing 3 months of work into 3 weeks.
Problem Statement & Engineering Motivation
Auditing over 600 legal merchant agreements (30 pages each) for mandatory compliance clauses and signature verifications was estimated at 3 full months of manual reading.
Technical Architecture & Implementation
Built a Python automation script using pdfplumber to extract text and tables deterministically, applied regex validation for standard clause formats, and routed ambiguous clauses to an LLM evaluator for semantic checking.
Key Technical Features
- Deterministic clause rule matching across 600+ documents
- Automated PDF discrepancy report generator
- Verification and cross-referencing against physical paper archives
- Automated flagging of complex commercial compliance terms
Technology & Component Stack
- Python 3.11
- pdfplumber
- PyPDF2
- Regular Expressions
- JSON Validation
- OCR
Outcomes & Engineering Lessons Learned
- A rule-based regex pipeline should handle 90% of deterministic checks before incurring LLM compute costs.
- Documenting edge-case rules upfront is the key to building resilient automation.
Related Research Study
This engineering build is connected with academic research investigation: /research/pdf-agreement-legal-automation