Finding a job is no easy task. It's even more difficult to find the right job. Too often, we think we’ve scored our dream job with the company we’ve long admired from afar, only to end up realizing it just wasn’t the right fit. There are so many factors to consider: the work culture, the hours, how much you need to travel---maybe even whether dogs are allowed in the office.
Cameron Levy and Dustin Smith, two former MIT students, were particularly interested in finding the right job, not only for themselves but for everyone. "Picking your job is among the most important decisions that you’ll ever make, and we couldn’t think of a more important problem to work on," Levy says. So in 2013, Levy, having finished his MBA, and Smith, who had just earned his computer science PhD, set out to build a tool that could better solve this problem.
Initial feedback was good. "When we started to tell our friends and classmates about our project, they all said the same thing: 'I need that.'" And then, in the fall of that year, the MIT Media Lab’s E14 Startup Fund---an incubator program for MIT graduates---agreed to back the project.
Today, little more than two years later, Levy and Smith are launching a website called Beansprock that uses natural language processing and machine learning techniques to match you with a suitable job. In other words, they're applying artificial intelligence to the career hunt. "It’s all about helping the job hunter find the best job in the job universe," Levy says.
The service is part of a massive movement towards artificial intelligence. Big tech companies, including Google, Facebook and Baidu, already use AI on the backend to power things like voice and image recognition, and countless smaller startups are getting into the game as well, looking to remake the length and breath of the business world---including job hunting.
To use the system, you feed it little bits of information about kinds of jobs you prefer. According to Levy, this takes about 2 to 5 minutes. You key in things like your preferred company size, information about what kind of culture you feel most comfortable in, skills you have or want to learn, and more.
If you like, you can also connect the service to your LinkedIn profile, feeding it more data. And you can tell it whether you’re a casual or serious job hunter, which can help define how often you'll get matches.
Meanwhile, on the other side of the equation, Beansprock pulls in thousands of new jobs every day from public listings and popular job sites. And, if needed, it augments the listing with additional information: what market that job is in, what the salary range might be, where the office location is, etc. By comparing this database with your personal info, the service's algorithms then match you with particular jobs.
It some cases it might ask you for additional information---say, whether you're interested in learning a particular skill. It can then use this info to find better matches. Plus, Levy says, its natural language algorithms are smart enough to understand complicated technical terms from the words that surround them. That means that if, say, a job seeker simply gave "ML" as a skill, Beansprock could understand that this means "machine learning."
At launch, Beansprock is limited to tech jobs, and listings are limited to Boston, New York, and San Francisco. The reason Levy and Smith decided to start with this sector, as they tell it, is that the IT skills gap is a very real thing---and they want to close it.
Beansprock does offer a few listings in sales and project management on its portal, but even those jobs are at tech companies. The bulk of the offerings, Levy says, are for IT professionals and software engineers. Levy and crew are also allowing business to feed their jobs listing directly into its system. When Beansprock makes a match, these businesses pay a fee.
Will the thing actually work? Will it scale? At least Levy says so: "We think it's going to get better and better with more and more data."
