Live application · Portfolio · Discuss a document workflow
A transparent document-review workspace for matching text against custom criteria, showing evidence, and exporting a structured result.
Recruiting and operations teams often review documents against the same checklist, but manual comparison is slow and opaque. This project demonstrates an inspectable first-pass workflow:
text input -> criteria normalization -> match + evidence -> review summary -> JSON export
- editable and
.txtdocument input; - custom, deduplicated criteria;
- deterministic match score with evidence excerpts;
- missing-criteria review and structured JSON export;
- responsive standalone website;
- dependency-free Node tests for edge cases.
The live score is deterministic decision support—not an AI model and never an automated hiring decision. The included documents are fictional.
npm test
python -m http.server 8000Open http://localhost:8000/site/.
A client version can add PDF/DOCX extraction, schema-validated AI outputs, batch processing, reviewer approval, role-based access, and retention controls behind a FastAPI service.
MIT — see LICENSE.