alphaspirit - Fotolia
AI vendors attack data scientist shortage with trainings
Internal data science training programs have helped vendors when colleges and universities have failed. Training is helping to fix the data scientist shortage.
The data scientist shortage, caused partially because AI and analytics technologies are constantly changing, and partially because colleges and universities can't keep up with the demand for data science education, is well-documented, and has been a problem for years.
To attack that shortage, data science vendors are developing internal and external training programs that provide education to their own employees and to their clients' employees.
Internal training programs
Fusemachines, Inc., a New York-based startup founded in 2013, provides AI experts, such as developers, engineers and programmers, to companies looking to build or refine their own AI systems.
Clients request all sorts of work, but right now, recommendation engines for retailers are in particularly high demand, said Steve Rennie, director of research at Fusemachines.
For its business model to work, Fusemachines needs high-quality data scientists. When it can't simply hire them, due to the data scientist shortage, it trains them.
"One of the cornerstones of our business structure is the internal education aspect," Rennie said.
Fusemachine's educational and training programs include a year-long AI Fellowship that takes in math and programming students and turns out machine learning engineers.
The program serves as training for participants, and also as a recruiting tool for Fusemachines. Participants are paired with in-house mentors and can participate in both on-site and online training.
Fusemachines works with students from around the world, Rennie said, including still-developing nations, and employs many data scientists from such nations.
"It's a great opportunity," Rennie said.
The company also provides training materials to clients that request them, to help clients get set up to work on their new AI systems.
Similarly, startup Cinnamon, an AI technology company that counts many new graduates among its few hundred internationally diverse employees, has recognized the data scientist shortage, and developed an internal training program.
Founded in 2012 in Tokyo, with offices around the world, the vendor created a two-week "boot camp" system to quickly teach new employees work-ready data science skills.
Training continues after the boot camp, with employees going to regular sessions and refresher courses. The programs are effective, according to Yoshiaki leda, co-founder and chief operating officer.
This type of internal training is relatively common for tech companies. Prominent analytics vendors such as Qlik and Tableau offer free or relatively low-cost training sessions. Clients can get employees working with new software fairly quickly. For vendors, the approach familiarizes customers with their products.
Cloudera, which sells data science, warehousing and management products that run on cloud or on-premises, regularly advises customers on their data science and AI projects.
"You can't just buy a machine learning (product), you have to learn how to adopt it," said Matt Brandwein, head of product for machine learning at Cloudera.
Cloudera also provides advisory services to customers about how to build products using Cloudera software, as well as for general knowledge about AI-related topics such as machine learning.
The vendor also provides education advisory services. These have seen an upsurge in customer demand recently, Brandwein said.
One customer, a United Kingdom-based statistics company, has been working with Cloudera on a project to onboard more statisticians, Brandwein said.
"We've been partnering with them on a training program, but in addition, giving them an environment where they can play around," he said.
While these vendor programs partially offset the data scientist shortage, the shortage is still there. Traditional colleges and universities have been unable to find enough money and staffing to keep up with the education demand themselves.
In the meantime, third-party education organizations have started to pop up.
Founded in 2014, The Data Incubator is one of these education providers. It works with more than 500 companies to help them train their data scientists.
"We take a data scientist and add ten percent to make them ready for the modern workforce," Michael Li, founder of The Data Incubator, said.
The New York-based company works with new graduates as well as more seasoned employees. Most clients are almost ready for the modern workforce, but are not quite there yet. New graduates might have a Ph.D. in mathematics, but no experience in machine learning. Older employees, meanwhile, might have much experience with machine learning, but not with using some of the newer machine learning tools.
The Data Incubator offers a few different types of training programs, including fellowships for students and corporate training sessions.
"What we've found is we're filling a void," Li said. "Universities are very theoretically driven. People in the industry are looking for that practical expertise."