Scaling Recruitment Through Automated Matching

Challenge

Intern-X is a global internship discovery and application platform. On its launch, it wanted to run a campaign reaching out to tens of thousands of University Students and Recent Graduates with some of the most appealing internships around the world. They reached out to Innovator with the desire to design and implement a scalable automatic matching model between talents and internship opportunities with the purpose of increasing user value and driving engagement. A perfect challenge for the left-brained part of our team.  

Solution

First and foremost, as the company had already implemented a profile creation wizard, we needed to get acquainted with the data collected. Once mapping all variables coming from user and internship profiles we worked with a team of HR experts to identify which signals matter when constructing a preference match. We then designed and implemented a multi-criteria decision model using ranked candidate as performance proxies, and ranked internship data as preference proxies. We then constructed a single index of relatedness between every internship and every applicant in the system. After having the basic model in place, we ran multiple simulations over the whole set of potential inputs to identify biases in the model, and potential excessive impact of certain characteristics. Once the model was balanced, we worked with a team of data engineers to implement it in production and increase the model's processing efficiency.  

Result

instead of providing a simplistic and semi-random recommendations for internships to apply to, we were able to give candidates a ranked order list of internships that they were likely to like and be suitable for. The result was a grand increase in profile completion rates and engagements (# post registration applications). Further details cannot be shared due to the sensitive nature of the project.