AI-powered career recommendation engine delivers more job options

A new cloud-based, online tool called MD Job Genie uses artificial intelligence to match employment seekers in Maryland with best-fit openings and training.

Launched last month, MD Job Genie builds on the state’s existing workforce services portal, the Maryland Workforce Exchange. To access the new tool, users can click its icon on the MWE webpage and enter information about themselves, including work and education history. Job seekers will get results that show openings in fields they have experience in—and some jobs they might not have considered but which the tool determined might be a good  fit based on data about successful job switches others have made.

“This really digs into those skills and that history, and then combines data—your resume, [unemployment insurance], wage data—in order to put together those matches,” said Joseph Farren, chief strategy officer at Maryland’s Labor Department (DOL), which oversees MWE and MD Job Genie. “It provides a deeper-level algorithm-based analysis of specific skills and job history to identify matches, not only in the field that you’re currently in, which a lot of engines do, but really what’s attractive about this tool is that it’s able to analyze different sectors and industries that are prioritizing your types of skills and job history.” 

To build MD Job Genie, DOL partnered with Research Improving People’s Lives (RIPL), a nonprofit that works with governments to help them use data, science and technology, and with Geographic Solutions, which provides the software that runs MWE.

“Job Genie is designed to feel like an enhancement to the Maryland Workforce Exchange,” said Scott Jensen, RIPL’s chief executive officer and vice president of external affairs. “That was something that was really important to the Labor Department. They didn’t want to have users feeling like there’s a million different and disconnected systems.” 

He summarized MD Job Genie as “sophisticated labor market statistics analysis.” RIPL used five years of unemployment insurance tax records to train the algorithm to make best-fit recommendations. Employers in Maryland report quarterly to the UI office what each employee who has a W2 tax form earns. They must indicate what their business is and what people—using Social Security numbers—work for them in what capacity.

“What we want to find is when somebody switches industries, did the switch work out?” Jensen said. “Was there a good causal impact on the switch? Was there a bad one? Or was it neutral? When we do that analysis, we do it on about 140,000 different jobs switches that people make … [so] that we can get an accurate prediction, an accurate analysis of the causal impact.”

Because that analysis involves a lot of sensitive data that MD Job Genie doesn’t need, RIPL and DOL also worked with the Maryland Department of Human Services’ MD THINK to de-identify it. MD THINK, which stands for the Maryland Total Human-services Integrated Network, is a cloud-based, shared-services platform and data repository designed to encourage information sharing. RIPL signed a security agreement with the state allowing it to use that de-identified data, but only in specific ways and only by certain people at the organization, Jensen said.

“To do our analysis … we don’t need to know who they are, we don't want to see their Social Security number,” Jensen said. RIPL only needs to know that the individual switched industries  and what the outcome was. MD THINK cleans that data, and then RIPL economists log on to a secure platform in MD THINK to conduct the analysis that results in a report showing industry switches and their causal impact. About a third end up being bad, a third neutral and a third successful, Jensen said. “Now when a person uses the Maryland Job Genie, they have the benefit of that analysis,” he said. “They are coming from a specific industry and if they want to switch industries, we can recommend to them things that they might want to look at.”

Users can save their results to return to later. They can also click through the results directly to job application pages or to sign up for training. 

“It’s really like Netflix for jobs, meant to just give you a quick card and say, ‘Hey you could work in this industry or that industry [and] this is how much they get paid. Click here and take a look at it,’” Jensen said. 

Other states use RIPL’s job search technology. It was incubated at the Rhode Island Department of Labor and Training and went live at the Hawaii Department of Labor and Industrial Relations last summer. More states, which Jensen said he could not yet name, will go live with it soon.

It took only two months to build MD Job Genie because the underlying framework was well established, but it still requires what Jensen called “significant tweaking” to integrate with each state’s ecosystem.

Although it’s too soon to assess MD Job Genie’s impact, Farren expects both Maryland residents and the state to see benefits. “When our customers are more successful in finding better, higher-paying jobs, the state is winning, not only because we’re saving resources, but when residents become more successful, the economy becomes more successful and everyone benefits.”

Stephanie Kanowitz is a freelance writer based in northern Virginia.