job description: If database administrators are jet pilots, then data modelers are aerospace engineers, performing the initial design work that enables production systems and applications to use data and drive business processes, according to David Van De Voort, principal human capital consultant with Mercer, a consulting firm. "They are in the business of creating data designs, defining relationships between data fields," he says. "They're doing the blueprint for a design which ultimately will get physically built on a machine."

why you need one: A business's data is one of its most important assets but it must be modeled efficiently to have great value. This is ever more important as corporate IT environments have become both more complex and more crucial to a business' success, says Taz Stephens, a regional managing director at IT services provider Technisource. "The importance of having an effective data modeling group has become increasingly important not only to the technology group but the business," he says. "To ensure that the applications and systems run as effectively as they can, you really need to have a good data modeler."

desired skills: At least five years of experience, familiarity with modeling tools such as ERWin. There is high demand for skills around data warehousing, business intelligence and ERP initiatives. Large companies want modelers with experience on more than one database. And match the modeler to the data environment. For example, companies with complex transactional systems running a terabyte or more of data want someone who has been in that world.

how to find one: Try job boards, word of mouth and other traditional means. However, the supply of candidates is tight. "There are user groups you really need to network through. This is not a post-and-pray kind of environment," said Jim Lanzalotto, vice president of strategy and marketing at talent and outsourcing firm Yoh.

what to look for: Data modelers need strong social skills since they interact with DBAs, developers, infrastructure resources and business users. Look for proactive thinking about how to use data in a way that aids the business. Candidates should have public speaking and presentation skills. elimination round: There is a sure-fire way to weed out the phonies, according to Stephens. If a data modeler can't demonstrate what they can do with an established modeling tool, "then they're a pretender," he says. Candidates who pass that test should then face questions meant to reveal how they might handle problem solving in your company's environment.