Data scientist Frank Lo has the type of problem most professionals can only dream of: He gets far too many job offers.
Lo, who leads a team of 16 data professionals at e-commerce website Wayfair, said he received about 150 inquiries from prospective employers last year. While he's not looking for a new gig, Lo said there is one thing above all else that he wants to know about a job.
"The first thing I ask them is, tell me about your data. Tell me about your problems," said Lo, who also founded datajobs.com. "You better have some pretty interesting problems for me to solve, or I'm not interested. I want to try to solve the unsolvable."
Data scientists are in high demand these days, but as employers scramble to recruit the best and brightest, the figures on their minds first and foremost are not necessarily salary and benefits, but data sets. The preference for positions that pit new data scientists against tough problems is biasing many candidates toward start-ups and small firms.
That could be a problem for companies seeking to build out big data teams in the coming years. Only 18 percent of companies are confident they have the talent to use data effectively, and just 19 percent say they are sure their data operations contribute to sales effectiveness, according to McKinsey & Co.
Employers seem to be willing to pay up for promising new hires. The median base salary for entry-level data scientists jumped 14 percent to $91,000 in the past year, according to a survey conducted by executive recruitment agency Burtch Works. That's the biggest income percent gain among all skill levels in the field.
Read More Big data's big stumbling block
While the term "data scientist" is now used broadly, true practitioners are capable of using both structured and unstructured data to predict outcomes and devise ways to take advantage of those predictions. They're not just making data visualizations.
To be sure, start-ups are scooping up a smaller percentage of these highly skilled data scientists as the discipline becomes increasingly mainstream. The proportion of data scientists employed by start-ups fell to 14 percent this March from 29 percent in 2014, Burtch Works reported.
However, there is a perception that entry-level data pros can tackle high-level work earlier in their career at fledging firms, professionals tell CNBC.