This repository documents and packages the data pipeline for the resume-parser-model project. The project explores fine-tuning compact language models to turn resume text into structured JSON, using only facts supported by each resume. The folders preserve the stages used to build the dataset: Resume files and their derived text/annotations use matching category folders and resume IDs so the processing trail can be followed across stages. Each JSONL record contains four top-level fields: - resumeid: the stable resume identifier shared across processing stages. - category: the resume's occupation/category label. - resumetext: the extracted text used as model input (X). - targetjson: the…