scripts/generate_resume_fixtures_eval.py

"""Additional resume fixtures for NER-chain evaluation.

Extends the original 4 fixtures in generate_resume_fixtures.py with 6 more
personas covering Backend, Data Engineer, AI Engineer (GenAI), DevOps/SRE,
BI Analyst, and ML Engineer — mapped against the 71-skill seeded catalog so
ground-truth canonical names line up 1:1 with Skill.skill_name.

Run: python backend/scripts/generate_resume_fixtures_eval.py
"""
from __future__ import annotations

from pathlib import Path

from reportlab.lib.pagesizes import LETTER
from reportlab.lib.styles import getSampleStyleSheet
from reportlab.platypus import Paragraph, SimpleDocTemplate, Spacer


BASE_DIR = Path(__file__).resolve().parent.parent
FIXTURES_DIR = BASE_DIR / "tests" / "fixtures" / "resumes"


def _write_paragraphs(path: Path, paragraphs: list[str]) -> None:
    styles = getSampleStyleSheet()
    doc = SimpleDocTemplate(
        str(path), pagesize=LETTER,
        leftMargin=54, rightMargin=54, topMargin=54, bottomMargin=54,
    )
    story = []
    for i, text in enumerate(paragraphs):
        style = styles["Heading2"] if i == 0 else styles["BodyText"]
        story.append(Paragraph(text, style))
        story.append(Spacer(1, 8))
    doc.build(story)


def resume_backend_dev() -> list[str]:
    return [
        "Hassan Raza - Software Engineering BS (UCP, Semester 7)",
        "<b>Summary:</b> Backend developer focused on Python and Django. "
        "Built production REST APIs for a logistics dashboard and a "
        "student course-registration portal. Comfortable with PostgreSQL "
        "schema design, Docker-based deployments, and Linux administration.",
        "<b>Skills:</b> Python (advanced), Django, Django REST Framework, "
        "PostgreSQL, Docker, Git, GitHub, Linux, REST API Design, JWT / OAuth, "
        "Bash Scripting.",
        "<b>Projects:</b> Internal billing API (Django + PostgreSQL, "
        "deployed with Docker Compose). Multi-tenant SaaS auth service "
        "using JWT token rotation. CI pipeline with GitHub Actions.",
        "<b>Internship - Backend Engineer Intern @ Systems Ltd (2025):</b> "
        "Optimised slow Django ORM queries, reduced p99 latency by 40%.",
        "<b>Education:</b> UCP Lahore, BS Software Engineering, CGPA 3.6/4.0.",
    ]


def resume_data_engineer() -> list[str]:
    return [
        "Sara Malik - BS Data Science (UCP, Semester 8)",
        "<b>Summary:</b> Data engineer with experience building batch and "
        "near-real-time pipelines. Lead author of an Airflow-orchestrated "
        "ETL stack that backs weekly marketing attribution reports.",
        "<b>Skills:</b> Python, SQL, Apache Spark, Apache Kafka, "
        "Apache Airflow, Snowflake, dbt, ETL / ELT, Docker, Git, "
        "Google BigQuery, PostgreSQL.",
        "<b>Projects:</b> Ingestion pipeline pulling click-stream events "
        "from Kafka into Snowflake via Spark Structured Streaming. "
        "dbt-based warehouse modelling with star-schema marts. Data "
        "quality tests with Great Expectations.",
        "<b>Internship - Data Engineer @ Careem (2025):</b> Authored 12 "
        "dbt models serving the rider-acquisition funnel.",
        "<b>Education:</b> UCP Lahore, BS Data Science, CGPA 3.85/4.0.",
    ]


def resume_ai_engineer_genai() -> list[str]:
    return [
        "Ahmad Tariq - BS Artificial Intelligence (UCP, Semester 7)",
        "<b>Summary:</b> AI engineer focused on LLM applications. "
        "Built retrieval-augmented chatbots using OpenAI and Anthropic "
        "models on internal corporate knowledge bases. Strong Python "
        "background, familiar with vector-DB retrieval patterns.",
        "<b>Skills:</b> Python, LangChain, LangGraph, OpenAI API, "
        "Anthropic Claude API, RAG, HuggingFace Transformers, "
        "Prompt Engineering, Sentence Transformers, pgvector, "
        "FastAPI, Git.",
        "<b>Projects:</b> Legal-document Q&A assistant using LangChain, "
        "OpenAI embeddings, and pgvector retrieval. Agentic research "
        "tool with LangGraph routing between Claude and GPT-4 per task.",
        "<b>Research:</b> Undergraduate thesis on prompt-ensembling for "
        "structured extraction tasks. Submitted workshop paper to a "
        "local NLP meetup.",
        "<b>Education:</b> UCP Lahore, BS Artificial Intelligence, "
        "CGPA 3.92/4.0.",
    ]


def resume_devops_sre() -> list[str]:
    return [
        "Umer Farooq - BS Computer Science (UCP, Semester 8)",
        "<b>Summary:</b> DevOps/SRE-track engineer. Manages production "
        "Kubernetes clusters for a fintech platform. Deep experience in "
        "infrastructure-as-code, container orchestration, and observability.",
        "<b>Skills:</b> Docker, Kubernetes, Ansible, Terraform, Prometheus, "
        "Grafana, CI/CD Pipelines, Jenkins, Linux, Bash Scripting, "
        "AWS, Git, GitHub Actions.",
        "<b>Projects:</b> Migrated a 30-service monolith from plain EC2 "
        "to EKS with Helm charts. Set up Prometheus + Grafana dashboards "
        "covering request-rate, error-rate, and saturation across services. "
        "Ansible-driven bootstrap for new clusters.",
        "<b>Internship - DevOps Engineer @ Tier5 (2025):</b> On-call rotation, "
        "authored 14 runbooks, reduced MTTR by 2x.",
        "<b>Education:</b> UCP Lahore, BS Computer Science, CGPA 3.7/4.0.",
    ]


def resume_bi_analyst() -> list[str]:
    return [
        "Mehwish Anwar - BS Data Science (UCP, Semester 6)",
        "<b>Summary:</b> BI analyst intern with experience in Power BI "
        "dashboards and SQL-based reporting. Comfortable with star-schema "
        "dimensional modelling and DAX measures.",
        "<b>Skills:</b> SQL, Microsoft Excel, Microsoft Power BI, Tableau, "
        "DAX, Power Query / M Language, Data Modeling, Data Storytelling, "
        "Star Schema Modeling, PostgreSQL.",
        "<b>Projects:</b> Executive-level sales dashboard on Power BI "
        "backed by a PostgreSQL star schema. Monthly churn-cohort Tableau "
        "workbook for a telecom case competition.",
        "<b>Internship - BI Analyst @ Telenor (2025):</b> Wrote 30+ SQL "
        "reports for the customer-experience team; rewrote 4 DAX measures "
        "that halved dashboard load time.",
        "<b>Education:</b> UCP Lahore, BS Data Science, CGPA 3.6/4.0.",
    ]


def resume_ml_engineer() -> list[str]:
    return [
        "Taimoor Abbas - BS Data Science (UCP, Semester 8)",
        "<b>Summary:</b> Machine learning engineer with strong deep-learning "
        "foundations. Shipped a computer-vision defect-detection model to "
        "production on a factory-floor edge device. Experienced in "
        "experiment-tracking and model-serving pipelines.",
        "<b>Skills:</b> Python, PyTorch, TensorFlow, Scikit-learn, MLflow, "
        "Docker, Feature Engineering, HuggingFace Transformers, NumPy, "
        "Pandas, Matplotlib, Git, Statistical Hypothesis Testing.",
        "<b>Projects:</b> PyTorch-based ResNet defect-classifier with "
        "MLflow run tracking. Dockerised model server with FastAPI. "
        "Sklearn baseline pipelines for a churn-prediction kaggle comp.",
        "<b>Internship - ML Engineer Intern @ Arbisoft (2025):</b> Built "
        "a feature store prototype; documented 18 features with lineage "
        "and freshness SLAs.",
        "<b>Education:</b> UCP Lahore, BS Data Science, CGPA 3.8/4.0.",
    ]


def main() -> None:
    FIXTURES_DIR.mkdir(parents=True, exist_ok=True)
    jobs = {
        "resume_backend_dev.pdf": resume_backend_dev(),
        "resume_data_engineer.pdf": resume_data_engineer(),
        "resume_ai_engineer_genai.pdf": resume_ai_engineer_genai(),
        "resume_devops_sre.pdf": resume_devops_sre(),
        "resume_bi_analyst.pdf": resume_bi_analyst(),
        "resume_ml_engineer.pdf": resume_ml_engineer(),
    }
    print(f"Generating eval fixtures into {FIXTURES_DIR}")
    for filename, paragraphs in jobs.items():
        out = FIXTURES_DIR / filename
        _write_paragraphs(out, paragraphs)
        print(f"  wrote {filename}  ({out.stat().st_size // 1024} KB)")
    print("Done.")


if __name__ == "__main__":
    main()