Built a natural language processing enabled smart search interface for ECCC specifically in aid of reducing some of the administrative overhead associated with selecting an appropriate generic job description for a job posting.

ECCC has provided roughly 1,000 generics that we use as our corpus and our team is currently building an AI model that can extract key terms and concepts from the data and return recommendations on the most appropriate generic to use based on high-level input from the end user.

The business case here is that for any new posting HR managers spend 5-7 days sifting through generics trying to find one that is relevant. We're aiming to give them a system that will do that in about 10 seconds.

The next phase is to build a resume ontology and start to leverage AI to evaluate candidates based on a particular job description. This is a much larger effort given the fact that resumes and candidate language is far more varied than the similarly structured generics.

We are planning to provide an interactive exploration tool that allows the end user to not only review candidate applications but adjust criteria on the fly and dynamically re-rank them based on the adjusted criteria. For example, if ECCC wanted to hire a CS-03 for a role that required supervision and they had specified 5 years of C# experience in the application but weren't finding the exact candidate they were after, using the interactive exploration tool they could play with the criteria and reduce it to 4 years of C# experience and see more candidates score near the top.

A final build would encompass a automated and intelligent job level classifier. The idea being that if you encounter a role where you need to create a unique, the new description can be submitted to the classifier to return a recommended level (e.g. CS-03).