We don’t want to be cheeky, but using machine learning is faster, cheaper, and better, in pretty much every possible way. Here’s what we mean.

Every “solution” on the resume filtering market involves forcing users to search through resumes by providing a list of specific set of keywords to find potential candidates to reach out to. Which means that if you want to hire someone with a degree from a top university (say, UC Berkeley), you actually have to tell your software explicitly to find every resume in your system with a degree from UC Berkeley.

That, of course, is only a partial solution. Searching for “UC Berkeley” will obviously give you a few great candidates, but you won’t see the candidates that are equally qualified from other great schools that you didn’t explicitly search for (like CalTech or Duke, for example). If you want to find those candidates, you’ll have to create scores of other searches, each naming a different top university, and hope you don’t forget to include any important university in your search.

It can get a bit messy.

And the problem only gets more complicated when you want to isolate other important factors, like a specific company in a candidate’s work history, or even just a job title, which can be incredibly difficult for keyword searching to differentiate. After all, just how many ways can people use different words to say “Software Engineer”?