Operationalizing AI: How Joveo Brings Artificial Intelligence to Life

These days it seems that artificial intelligence (AI) technology is powering everything from robots to toasters. AI figures prominently into Joveo’s solution portfolio, as well. We recently sat down with Rohan Parakh, Group Product Manager – Data Science, to learn how Joveo operationalizes artificial intelligence in recruitment processes. Through this process, Joveo’s data science and product teams transform potential innovations into highly-performing analytic features that deliver highly predictive, unbiased decisions. Here’s a recap of our conversation.

From Harvard to Joveo

Rohan has been keenly interested in applying analytics to help job seekers for a half-dozen years; when he was a graduate student at the Harvard Kennedy School, he developed Two Cents, a system that crowdsources referrals from peers and supervisors to create a reputation score for candidates. “The goal of the platform was to make identifying talent easier for employers, and to match candidates to jobs they are best suited for,” Rohan said.

Today at Joveo, Rohan is putting his talent to work in product management, functioning as a critical conduit between the business and Joveo’s team of data scientists. He said, “My role is identifying the right business problems that need data science intervention – opportunities where AI can make a difference. I work with the product management organization to prioritize those opportunities, figure out which will have the greatest business impact, and then work with our data science team on execution.”

Four inputs for AI innovation

Rohan walked us through the multiple channels through which business ideas for AI innovation flow. “We broadly rely on four sources for requests,” he said.

• Product teams: Joveo’s product management teams are constantly identifying prospective analytic enhancements, which they present to Rohan. “Our product teams also are in frequent contact with customers, who make requests directly. In turn, these requests are processed and prioritized.”
• Other internal customers: “Joveo’s data science product team cuts across multiple areas of our organization,” Rohan explained. “It’s similar to a design team being called upon by other parts of the organization.” Engineering, executive management and product marketing groups can all weigh in with business problems that can potentially be addressed with AI in Joveo’s solutions.
• Large customers: Many client organizations are big users of the algorithms currently embedded in Joveo solutions. From their day in, day out experience in using these products, customers contact Joveo directly with business needs that can potentially be met through AI enhancements. 
• Engagement workshops: Joveo holds periodic ideation workshops with customers and internal product experts. These highly creative gatherings “generate a lot of ideas based on customer pain points,” Rohan says. “We work through these pain points very intentionally, looking at how AI may be applied to address customer issues.”

Liaising with data scientists

Rohan continually casts a wide innovation net, gathering ideas for AI enhancements. But not every potential improvement makes sense, and “some problems can’t be solved by data science or AI.” He provides context: “Platform product managers typically talk with users and understand their pain points. From those pain points we distill out problem statements, and determine where data science can help. But since our product managers are not data scientists, they might not always suggest problems that are solvable, even by the most sophisticated AI.”

Joveo’s data scientists provide decisive insight at this stage. “They are the ones who can tell if we have the right data, and if it can be extended to address the new problem set,” Rohan says. “The process is a bit iterative, but ultimately it’s much more time-effective. We try to reduce unnecessary time spent building models by first validating whether any proposed enhancement is something Joveo can and would do.” 

AI innovation in action

Although perhaps not completely linear, Joveo’s approach to developing AI is highly effective. “At every point across the talent acquisition journey, decisions must be made. We don’t want decision making to be guesswork or subject to trial and error,” Rohan says. Joveo uses AI extensively to automate processes and decisioning, reducing the need for human intervention and thus bias and potential errors. 

Rohan provided context for discussing AI by describing the essence of Joveo customers’ programmatic job advertising needs: “When a job comes to our system, our clients essentially want to identify the most effective places to publish these jobs for the least money. They have tens of thousands of options and, without Joveo, they rely on expedience, gut instinct and guesswork. While it’s possible to meet ad performance goals with this approach some of the time, it’s generally not sustainable.” He then described three key areas as examples of AI-driven automation that help Joveo customers achieve their goals today:

• Personification: “The goal is for the machine [the Joveo platform] is to be able to understand the ad and perform the client’s job,” Rohan said, republishing it on the most effective job boards for the lowest cost. For example, let’s say Joveo receives an erroneous job ad for a traveling nurse in Boston, San Francisco, USA. We humans understand that the ad is referring to Boston, Mass. But AI technology is what figures out that the job is in Boston, Mass.” Furthermore, AI also enables the machine to understand the job, classifying which occupation it belongs to, what location it belongs to, the company, and the industry.“Now that we know what the job is, we should be able to determine what capabilities candidates need, and make decisions about that,” he continued. “These are the building blocks.”
• Optimization: Rohan’s second AI example was a bit easier to digest. Optimization powers Joveo’s media planning solution, determining where to place ads at the lowest cost. AI drives Joveo’s bid intelligence to determine the optimal ad bid at any given moment. Here, optimization allows Joveo clients to “get the best performance for the ad at the lowest cost,” Rohan said, helping to guide where media budgets should be spent.A second major application for optimization is candidate relevance. Joveo uses AI technology to precision-sort thousands of resumes that a single job ad posting may attract, alleviating recruiters from a tedious manual process.
• Expansions: Rohan explained the concept of expansions: If I post a job that’s located in New York, I may miss candidates from New Jersey. But if I don’t post for New Jersey, it will never show up in their search. With expansions, Joveo creates copies of the job for nearby locations, for example, to all the ZIP codes within a 10- or 20-mile radius of where the job is located, to reach more potential applicants.” Further, he noted that Joveo can additionally expand job titles, such as from ‘Project Manager’ to ‘Program Manager,’ to boost conversion rates.

For a deeper dive into Joveo’s data science team, and more discussion of personification, optimization and expansions, check out Rohan’s blog, “Joveo’s Data Science Team Changes the World One Job at a Time.” 

Learn more about how our global team can give you total picture of recruitment advertising costs by requesting a demo of Joveo solutions today. And follow us on Twitter and LinkedIn, where we’re always working to help you get the most out of your recruitment advertising.