Data scientists are a group of highly educated people; highly valued and increasingly important in the modern world, many of them are lucky enough to be able to pick and choose where they work. An organisation dedicated to improving the data science job market, then, might not appear at first glance to be the most obvious business strategy.
But for the founders of Brainpool.ai, their experiences demonstrated to them that it was not just an opportunity, but a necessity. Kasia Borowska, managing director, Dr Paula Parpart, head of research and Dr Peter Bebbington, CTO, are on a mission to improve the data science job market.
Data science and data scientists always develop in academia, in the vaunted halls of some of the most prestigious universities in the world. Groundbreaking projects, innovative research and developments with the power to change the world have come out of academic data science departments.
But, the company’s founders discovered, there exists a big gap between these academic pursuits and the industries that could adopt and commercialise them. Businesses have typically struggled to access the type of research that could be helpful to them, and academia has found commercialisation difficult.
The Brainpool mission
Brainpool was founded with the intention of bridging this gap. “The problem that most top-level data scientists have is a lot of recruitment emails,” says Borowska.
Data scientists are dynamic individuals – they want to be working on the newest and best data sets using state of the art technologies
“There’s a 50% gap between supply and demand. This means that data scientists can pick and choose projects – and they’re usually not interested in nine to five corporation jobs. So, what we’re trying to do is enable clients to easily find academics that match their requirements and allow scientists to find the interesting projects.”
What’s caused the gap between academic data scientists and industry? Do data scientists simply not fit in a commercial environment? Borowska believes that some of the qualities inherent to data scientists aren’t compatible with a typical job.
“There’s the repetition of working for one company. People stop learning when they go to a corporate nine to five job; doing the same tasks each day means they progress slower in terms of their development, compared to academia.”
“That’s a polite way of saying that data scientists have short attention spans,” laughs Bebbington. “They’re quite dynamic individuals – they want to be working on the newest and best data sets and using state of the art technologies.
“There has to be a revolving door between academia and private enterprise. It’s such a fast-moving area – one week a technology comes out, the next it’s redundant. There has to be direct access to academia.”
Founded out of UCL in February 2016, the company now has seven employees and a network of around 300 data scientists involved in machine learning and AI. These aren’t recruitment consultants – its founders are academics, have worked, and continue to work, in the sector. Paula Parpart, one of the company’s founders, though still involved, is back focussing on a post-doctorate qualification in true AI.
This gives the team a unique perspective on some of the developments, fear and hype involved in both the commercial and academic applications of machine learning and AI. It’s now familiar ground for most people with a passing interest in the area, but misconceptions over the difference between machine learning, artificial intelligence and artificial general intelligence still remain. It’s confusions like these that means the industry needs to be nurtured by people with an understanding of the topic.
The democratisation of data
Another of the hot topics in the area of artificial intelligence, in particular, is the democratisation of data, models, and the people involved in developing AI. Abraham Gilbert, head of ML at Brainpool, argues that keeping AI open is a major consideration.
“There’s an AI arms race to amass talent and become the next superpower. The AI community is so enthralled by the science in this age of discovery that it hasn’t yet stopped to examine the risks from any individual company, government or person controlling so much of the world’s intellectual stock.
“That’s why democratisation of access to AI is so important because it enables the positives of AI discovery to come to the fore and prevents either monopolies forming or checks and balances against companies or governments misusing their power. That’s what’s so attractive about working for and with Brainpool.ai because its model can give AI access to the masses.”
On the topic of ethics, another major fear that bubbles up in any conversation about AI is job loss. There are many clichés bandied about on this topic, in particular with reference to the Industrial Revolution and cars replacing horses, but those in the know say that for now, the matter is a lot simpler.
“Sensation sells,” Bebbington tells me. “At the rate that we’re going, it will be a lot more about augmentation, taking away the mundane aspects of the job. People will be able to focus on the creative side of things – that’s the current trajectory.
“Maybe 100 years in the future there’ll be aspects of machines doing a lot of jobs and humans doing creative aspects. At this stage, it’s nothing to worry about.”
The network effect
What the team at Brainpool does focus on is the quality of its network, and helping them get the most out of their careers and helping its clients get the most of them. According to Borowska, they are careful to only take the cream of the crop, people who are in a position to choose jobs. And that is a vital change. It used to be that a job would have thousands of applicants, but these days, a company has to work very hard to find suitable applicants at all. The ball is back in their court.
And with its network of experts, Brainpool has an advantage. Some commercial deals are emerging; one of the first in the pipeline. Though the details are currently confidential, it’s a project working on the development of a product that uses deep learning aided algorithms to predict interest rate futures.
That’s in exploration status at the moment, but Borowska hopes to have a first finished product by the end of the year. The aim, she says, is to scale that side of the business, got more data scientists involved, and move towards an incubator model. With the UK AI market developing as it is, it could prove lucrative.