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.
About a third of graduates from Northwestern University's Master of Science program in analytics strongly prefer to work for small firms, while another third are focused on finding employment in the Chicago area, and the rest gravitate toward consultancies and financial firms, said Diego Klabjan, director of the program and a partner at Opex Analytics. Meanwhile, few grads show a strong desire to work for multinationals, he said.
"Some of them go start working for bigger companies, and then they find out that there's not much innovation in those large companies, so they are essentially doing [business intelligence] stuff, but their passion is really data science," he said.
Read MoreWhy your kids will want to be data scientists
Matt Yancey, 29, said he was actively seeking employment at a start-up before graduating from Northwestern in December. He began his career at Accenture, a multinational consulting and outplacement firm, and said he wanted to find an employer that would let him use the broad range of skills he learned as a student.
For Yancey, that was First Analytics, an analytics consultancy and software development firm with just more than a dozen data scientists founded in 2009.
"If you're going into a smaller company, chances are they don't already have a bunch of processes already built in place," he told CNBC. "If I were to go work for some Fortune 500 company I'd have to be programming in whatever language they wanted or use whatever tools they wanted."
At the same time, Yancey has eschewed the Silicon Valley career path, saying his cost of living in Sugar Grove, Illinois, is far lower. He didn't want to get stuck into a shoebox-sized apartment in San Francisco or endure a long daily commute from a more affordable area.
One of Yancey's former classmates, Andrew Fox, 28, ended up going to work at Opex Analytics for Klabjan. He worked at Disney and Nielsen before changing career tracks, and said he too worried his skill set would be underutilized at a large company.
"You don't know if the leadership at all levels gets advanced analytics," he said. "On the other hand, the three founders at Opex knew data science very, very well."
Still, many data professionals have now put in time at start-ups that ultimately failed or have seen colleagues' companies go under, said Linda Burtch, managing director at Burtch Works. For these people, working for an established firm—which may offer better work-life balance and a superior benefit package—is increasingly attractive.
"The shine of the start-up is starting to wear off a little bit," Burtch said.
Read MoreWhy analyzing Big Data can be bad for business
Alice Zhao, 27 and a graduate of Northwestern's first class of data scientists in 2013, said work-life balance was a major factor in her decision to work at Cars.com, the world's second-largest classified site for autos.
Zhao was also the first data scientist at Cars.com and said the opportunity to make a mark was a big draw. Further, she said she'd seen colleagues at start-ups get roped into performing menial tasks such as data warehousing and data cleaning.
"I feel pretty lucky here. I get some direction from my director, but I've also had a lot of freedom to explore data sets," she said. Her schedule also affords her time to work on her own data projects.
The young data professionals are not necessarily opposed to working at legacy companies in the future, but they say they'd prefer to be at the helm of their own team of scientists within those organizations.
Companies looking to recruit entry-level candidates could learn a thing or two from start-ups, Burtch said. While a small company might pull the trigger after a Skype interview, she explained, large corporations stand to lose potential employees during long, multistep hiring processes.
"There's a lot of problems with legacy companies because they think they can take a legacy way of attracting candidates," Burtch said. "There's a huge difference between the way they recruit, and the advantage goes to the more nimble organization."