Apziva Project 3 - HR Talent Search

Overview

This project is designed to facilitate HR talent search processes. It leverages data-driven methodologies to enhance the recruitment workflow, aiming to match candidates effectively with keywords using Natural Language Processing (NLP) and Large Language Models (LLMs).

Dataset

The data comes from our sourcing efforts. We removed any field that could directly reveal personal details and gave a unique identifier for each candidate.

id : unique identifier for candidate (numeric)
job_title : job title for candidate (text)
location : geographical location for candidate (text)
connections: number of connections candidate has, 500+ means over 500 (text)

Project Structure

The repository follows a structured layout to organize code, data, and documentation effectively:

    ├── README.md           <- Project overview and instructions.
    ├── apzivaproject3      <- All the Scripts Used to Generate the Results
    │   ├── classes         <- Any custom classes I wrote goes here
    │   ├── dataset         <- Scripts related to cleaning the inital data
    │   ├── functions       <- Scripts containing standalone functions
    │   ├── modeling        <- All model training and predictions
    │   ├── setup           <- Scripts for additional cleaning and feature creation
    ├── docs                <- Project documentation.
    ├── notebooks           <- Jupyter notebooks for exploration and analysis.
    ├── references          <- Reference materials and related resources.
    ├── reports             <- Generated reports and visualizations.
    ├── environment.yml     <- Conda environment specifications.
    ├── requirements.txt    <- Python package dependencies.
    ├── setup.cfg           <- Configuration for package distribution.
    ├── pyproject.toml      <- Build system requirements.
    ├── main.py             <- Main execution script.

Acknowledgments

Apziva for providing the dataset and project framework.​ The open-source community for their invaluable tools and libraries.​