Candidate Scoring Models for a Job Search Platform

Summary

• While there are numerous job boards, few of them vet and score candidates based on their learning skills and actual technical skills.
• dhired.com aims to do just that. It is a niche job search portal for searching top technical talent integrated with an online & offline schools, served as a mechanism for filtering candidates.
• DataRoot Labs built the platform from end-to-end including an AI components which filter and score the candidates.

Tech Stack

Timeline

Tech Challenge

• The platform has a large number of students that become job candidates after finishing the online school. The main challenge lies in sorting students during the studying cycle to identify best talent.
• Another challenge is to supply students with personalized study program. The program provides curated content such as customized tests and materials based on students' skills level and needs.

Solution

• We created a series of scoring models based on the datasets and additionally publicly available data on prior students’ activity.
• Those models evaluate candidates upon their registration for the online school, recommend the most effective studying roadmaps for a given subject.
• Additionally, via push notifications and recommendations of additional studying material the models optimize the collaboration process to recommend the most efficient learning process.

Impact

• The LTV of student cohorts doubled from 5% to 10%, based on the number of students who finished the school.
• The tracking of "student-content' interaction allowed for aggregating more data and automatically enriching their internally created CVs that are later passed to potential employers.
• As a result, students' probability to get the desired employment significantly increases. At the same time, employers get candidates that are more qualified for the position.