hr-oxy-poc / README.md

AI-Powered Automated CV Screening for OXY HR

Last updated: 4/16/2026GitHub

AI-Powered Automated CV Screening for OXY HR

Full Solution Scope (Post-POC Vision)

The full solution will integrate directly with Workday, OXY's HR system, and include:

  1. Automated CV Ingestion

    • Applications are automatically received in Workday.
    • Siraaj Agent retrieves and queues new applications.
  2. AI-Powered Screening

    • Siraaj Agent evaluates each CV using role-specific criteria, rubrics, and requirements provided by OXY HR.
    • Generates a scoring summary for each candidate.
  3. Recommendation & Feedback Loop

    • Posts a report and recommendation back to Workday.
    • Optionally auto-approves or auto-rejects candidates based on HR-configured thresholds.
  4. Continuous Improvement

    • HR can provide feedback on AI evaluations to refine accuracy over time.

POC Scope

The POC will be a simplified demonstration of the above process, focused on proving feasibility and value.

In-Scope Activities:

  • Test CVs will be uploaded through a simple web interface.
  • A sample job description and screening criteria will be provided by OXY HR.
  • Siraaj Agent will:
    1. Analyze each test CV against the job description and criteria.
    2. Generate a summary report and screening recommendation.
    3. Provide an output report viewable via the web interface.

Out-of-Scope (For POC):

  • Direct integration with Workday.
  • Automatic approval/rejection in production systems.
  • Screening of live/real applicant data (only test data will be used).

Problem Statement

  • Manual CV screening is time-consuming and repetitive.
  • It creates bottlenecks in the recruitment process.
  • There is inconsistency in evaluations due to human subjectivity.
  • Scaling up recruitment during peak hiring periods is challenging.

Supporting documents

N/A

Getting started

Pre-requisites

Project setup

Clone the repository

git clone git@github.com:rihal-om/hr-oxy-poc.git
cd hr-oxy-poc

Running the app locally

  1. Create and fill .env

  2. Run docker-compose to start the application

docker-compose up
  1. Once the application is running, open your browser and navigate to:
# FastAPI Endpoints
http://localhost:8000/docs
# Streamlit Frontend
http://localhost:8501

Tech Stack

CategoryToolDescription
FrontendStreamlit UIUI library/framework
BackendFastAPI + LangGraphServer-side language/framework
Containerization{{DOCKER}}Container platform

Project Team

NameRoleGitHub
Bushra Al JahwariMachine Learning Engineer@bushrxh

Feel free to reach out to any of us for questions, feedback, or collaboration opportunities. We are always open to discussing new ideas and improvements for the project.

Happy coding!