abhaykanjoor/iteratecv-resume-tailoring

IterateCV Resume Tailoring Dataset

Dataset Description

An instruction-tuning dataset for fine-tuning LLMs to tailor resumes to job descriptions. Built for the IterateCV project.

Each example contains:

• —instruction: Task description for the model
• —input: A master resume in JSON format + a job description
• —output: A tailored version of the resume optimized for the job description

Dataset Construction

1. 1.Source resume-JD pairs from `cnamuangtoun/resume-job-description-fit` (Good Fit + Potential Fit labels only)
2. 2.Parsed raw resume text into structured JSON using GPT-4o mini with a custom Pydantic schema (MasterResume)
3. 3.Generated tailored resume outputs using GPT-4o mini with a carefully engineered tailoring prompt
4. 4.Formatted into Alpaca-style instruction-tuning JSONL

Schema

The resume JSON follows this schema:

personal_info: {name, email, phone, location, links}
summary: str
experience: [{title, company, dates, bullet_points}]
projects: [{title, tech_stack, description, bullet_points}]
skills: {languages, frameworks, tools, domain_skills}
education: [{degree, institution, dates}]
awards: [{title, year, issuer, description}]
leadership: [{role, organization, dates}]

Splits

Usage

from datasets import load_dataset

ds = load_dataset("abhaykanjoor/iteratecv-resume-tailoring")
print(ds["train"][0])

Fine-tuned Model

This dataset was used to fine-tune Llama 3.2-3B using QLoRA. Model available at: abhaykanjoor/iteratecv-llama-3.2-3b-lora (coming soon)

License

MIT