JOBMIrror - Job Application Success Prediction

Problem Statement

In 2025, mass application for jobs can be tiresome and ineffective for the average individual since on average an applicant has to apply to more than 100 applications. Managing each individual application at that scale, where a person’s resume needs to be catered to each different job posting is unsustainable and can lead to application fatigue and burnout.

Solution

My group proposes a simple machine learning model to predict whether based on your resume / job application and the job description, you would get an interview. LLMs were leveraged to extract details from both the resume and job description and Airflow + Mlflow + EvidentlyAI was used to schedule and keep track of model metrics to observe model performance over time