Will Your Resumé Impress the Computer?

Burning Glass Technologies says its new artificial intelligence-based software can separate the stars from the slackers

Wondering why you never got a callback on your killer resumé? Perhaps a new resumé screener that companies are using pegged you as a job-jumper, when the folks who are making the hiring decision want a lifer. Software that helps recruiters match skills listed on resumés with job openings has been around for awhile. A private San Diego company, however, claims to be the first to use artificial intelligence to help companies predict not only whether a prospective hire is a good fit for the job but also whether the candidate will stick around.

Founded in October, 1999, by several scientists who developed sophisticated software used to detect credit-card fraud, Burning Glass Technologies (taken from the old name for a lens used to focus the sun's rays to create heat) says four recruiting-services companies -- Tapestry.Net, Guru, Sendouts.com, and National Search Associates -- have bought its product, called Lens. Through these clients' partners, more than 1,000 recruiters and companies actually use it, says Krishna Gopinathan, the company's chief executive and co-founder.

In fact, people who have applied to such biggies as Microsoft, Sun Microsystems, and Southwest Airlines may have this new HAL of the HR world to blame for getting their resumés booted from the pool. ("I'm sorry, Dave. I'm afraid I can't interview you.") These companies have all turned over resumés to recruiters that have filtered them using Burning Glass' technology, Gopinathan says.


  Lens uses statistical models to predict a job-seeker's aptitude for a position, future performance, ability to manage, and other behavior based on the self-described work history of previous individuals as recorded in Burning Glass's databank, which contains 200,000 resumés and counting, Gopinathan says.

Elements it looks at include a prospective hire's past employers, job titles, skills learned, education, and general career path. (The system keeps getting "smarter" as Burning Glass feeds into the database each fresh resumé that its recruiting industry customers scan with the technology.) resumés are "blinded" -- the candidate's name and contact information are removed -- so Burning Glass avoids privacy problems, Gopinanthan says.

Though Burning Glass predicts annual revenues of only $5 million to $10 million, and expects to lose $4 million this year, it's relying on advanced technology to break into a $500 million segment of the recruiting business that's growing 60% a year. Less advanced resumé screeners stick to keyword or field searches to flag candidates with, let's say, C++ programming experience.


  Lens, however, which costs $200,000 to $1 million a year to clients depending on the volume of resumés scanned, might "learn" that people who start their careers as software engineers tend to become senior software engineers before rising to project manager. And it calculates the time it took for each promotion.

The program also makes a prediction on employee-company fit. It might conclude that consultants from Cap Gemini Ernst & Young tend to stay for awhile if they go work for KPMG and vice versa, Gopinathan says.

"Basically, it's able to look at how long similar candidates have held similar positions in similar industries," Gopinanthan adds. "We can train it to predict the career path of an individual. If it figures out that someone is a fast-track person, this may be a signal to the hiring company that the person needs to be promoted quickly if the company wants to keep them."


  Isn't this something any experienced hirer should be able to figure out? Well, yes, says, Gopinathan, except that a human being can only look at several dozen resumés in an hour -- during which Lens can sift through millions. "We don't replace the recruiter," Gopinathan says. "But you can use artificial intelligence and bring the right people to the attention of the human recruiter."

Gopinathan says Lens also recognizes that job-seekers often tend to pad their resumés. For instance, if there is an inconsistency between a job title and a job description, it might down-weight the importance of that job in its modeling prediction. It's too early to tell just how on target Lens's predictions are, Gopinathan says. The company began shipping its product on Apr. 30.

Dayle Bowen, CEO of Tapestry.Net, a Scotts Valley (Calif). company that provides job applicants to recruiters, among other services, is a fan of Burning Glass's technology. He says Lens has greatly reduced the time it takes to find quality candidates. "People have been matching skills with jobs ever since databases existed," Bowen says. "But this can read and understand the context and predict the likelihood someone will succeed in a job."

Still, Bowen is skeptical of Lens' ability to predict employee tenure. "Ultimately," he says, "longevity seems to me like it's more of a cultural consideration that has to be taken care of by a human being." Which means that, like the humans who work with people, Lens can never afford to stop learning.

By Eric Wahlgren in New York

Before it's here, it's on the Bloomberg Terminal.