In this case study, we will look at a Natural Language Processing aka NLP-enabled pipeline built to improve candidate & contractor sourcing inside a private database. The client had complex problem in automating their hiring pipeline and their organization was wasting too much time in hiring employees and contractors. They required a software-aided way to quickly sort through thousands of applications to prioritize candidates previously employed by the organization and those having skills similar to them.
This product pipeline was built for one of the world’s largest employers. The business value was centered around a robust system capable of sourcing previously vetted/employed candidates as well as lookalikes across several private & public databases given an input of a resume or equivalent information.
A private database was provided by the client for the creation of look alike profiles of candidates. A specific subset of the data was chosen to be used for external validation as well as black-box testing. The database was fractured across multiple instances & regions, a centralized scheme was proposed and subsequently developed to capture the collated database in its entirety – reaching a total of million+ records. The training data was extracted from the records and a persona list was created to curate necessary subsets for training & validation. An architecture was proposed involving a scalable Progressive Web App (PWA) was proposed to be used across desktops and mobile devices while a backend consisting of a python based API service would be used to handle the NLP and data processing modules.
The development cycle lasted a total of six months across fourteen sprints. The first useable module was provided to HR by the end of the first month into development. Multiple layers of detection and classification were built to enable a robust similarity engine capable of sorting results based on likeness to the supplied sample resume. Further, the engine was modularized to create a fully customizable search filter that allows the user to sort the results in any desired fashion. Multiple data integrity challenges were tackled successfully by the team and demonstrated that even the great wild west of unstructured resume data could be conquered with the right plan and engineer team.
The models were built into a scalable API that could be deployed on the organization’s private cloud in AWS. A full-scale frontend application was designed and built to the client’s requirements that satisfied u/x requirements and cut short candidate searches by as much as 800%. The organization saw an immediate return on investment (RoI) across the deployed job roles and plans to expand to more job roles & hiring functions in the future. AI based sorting and list reduction is a huge boost to any hiring department and can revolutionize even the best of the organizations. Consider using AI in your next hiring pipeline!