Top 5 takeaways from the WeAreDevelopers webinar ‘How to Recruit a Data Scientist’

In a wide-ranging webinar on How to Recruit a Data Scientist, three speakers joined WeAreDevelopers managing director Rudi Bauer.

The trio; Hung Lee of Recruiting Brainfood, Yves Greijn of Maven, and Jodie Burchell, a data scientist at JetBrains, said their goal was simple.

The aim was to banish “misunderstandings” and “stereotypes” of two separate “worlds” -- Recruitment and Data Science, said Bauer.

But even to describe the March 14th online event as a “webinar” risks adding to the misunderstanding.

“It’s a conversation, not a webinar,” clarified Lee, a few minutes in, referring to LinkedIn users (still) being able to add their own views about the online session’s talking points – of which the five biggest were as follows.

Top 5 takeaways from the WeAreDevelopers ‘conversation’ How to Recruit a Data Scientist

1: ChatGPT pushed down the Data Scientist hiring accelerator

“Since the game-changing move OpenAI did by releasing ChatGPT, it’s pushed the accelerator down on how many of these people… [skilled in Data Science] need to be recruited”, said Lee.

The founder of Recruiting Brainfood, Lee believes that the November 2022 launch of ChatGPT increased the “noise level” and “activity” around Data Science.

But more tangibly, the popular AI tool has also “increased the encounters recruiters [now] have with Data Science” candidates.

'You need to probably have a data scientist in your team as well -- because everybody has got one'

Later, at almost the 20 minute-mark, as to whether companies are describing the “project and problem” that they need a data scientist for well enough (the suggestion was that they are not), Bauer offered:

“Hiring a data scientist is something everybody tries to do…[in 2024] because it’s a little bit unique to use AI at the moment.

“So you need to probably have a data scientist in your team as well -- [just] because everybody [else] has got one!”

It would be better if companies first said to themselves ‘we need to understand the problem; and we need to describe the problem.’

Only then, continued Bauer, can companies begin to find “the right team, and the right solution.”

2: Data Engineering isn’t Data Science’s poorer cousin

“Sometimes we’re seeing that the market is pushing people with a technical background, for instance a Computer Science or Computer Engineering degree, into Data Science -- because that’s where all the funky stuff happens,” said Maven’s Yves Greijn, putting air quotes around ‘funky stuff.’

“Yet solid data engineers, who are senior, and who manage to stay on the data engineering side, are incredibly rare to find -- and have so much value,” Greijn said.

The founder of Maven then directly addressed technical candidates who are weighing up their options:

“Don’t think you necessarily need to be on the data science side, to do very well for yourself.”

'You need between two and five data engineers for every data scientist'

Nodding in agreement to Greijn’s show of support for data engineers, was Jodie Burchell.

A data scientist for the last eight years who had an academic career before joining an ad-tech firm as its lead data scientist, Burchell modestly mused that she wouldn’t get past stage one of a data engineering interview.

“I don’t want to devalue my own work,” she cautioned with a smile, “but depending on the size and complexity of your data, I would say you need between two and five data engineers for every data scientist.”

Burchell hinted some companies wrongly overlook the importance of having a strong team of data engineers to “manage living data as it comes in; store it in a way that’s accessible and store it in a way that’s affordable.”

3: The Rolls-Royce of solutions is NOT always necessary

About a quarter into the webinar, Lee asked if recruiters of data analysts, data engineers or data scientists ever get so perplexed that they’d like to ask candidates; ‘Why are you doing that?!’

Maven’s Greijn answered almost instantly -- and seemingly with contractors in mind: “Overcomplicating solutions.

“Yes, there’s so many things you can model. You could build a deep neural network. But the question is – ‘Is your client going to be satisfied?’

“Even for very senior roles [when we interview-test candidates when we’re recruiting them], I sometimes ask myself… [of their answers], ‘Is this just overcomplicating the solution for [the sake of] building something fancy?’”

'A fixed solution to a set problem that makes sense in the real-world'

The Maven recruiter said the tendency to overcomplicate is “often” the first negative he notices with data science candidates.

Making clear he was speaking about his personal preference, (but Bauer, Lee and Burchell all nodded along to it), Greijn added: “I’m impressed with a fixed solution to a set problem that makes sense in the real-world.”

Lee warned candidates with a penchant for overcomplicating: “It might well be worth bearing in mind [then, with] some recruiters at least, [they] may develop the perception that you’re impractical.”

'Show how you solved a problem where there was a significant restraint that created a sub-optimal outcome'

Recruiting Brainfood’s boss recommends that candidates should strive to do the opposite -- actually demonstrate their practicality.

And to demonstrate practicality, Lee urged candidates to bring up from their past a problem (which they solved) where there was a “significant restraint that created a sub-optimal outcome.”

When doing some hiring of her own, data scientist Burchell said a straightforward answer -- Linear Regression -- was what she was hoping to hear from data science job-hopefuls, when she quizzed: ‘Which is the first model you’d build and why?’

She added: “It’s about scoping. It’s about doing the minimal solution. It’s also about disabusing yourself of the notion that you cannot timebox research.”

'On your CV as a Data Scientist, bullet-point things that show the commercial value you created'

Burchell also said solutions that fit the problem should be concisely flagged up by data scientist contractors on their CVs, assuming getting hired is the goal.

“Bullet-point things where you can show the commercial value [you created],” she said.

 “Like if you can say, ‘I increased the customer acquisition by 500 people per month’ or if you can say ‘I saved the company 10%.’ These are amazing selling points, because you show your value.”

Keen not to lose the point about overcomplication, the Maven recruiter, Greijn, reinforced to data scientist candidates heading for a technical interview test, or trying to impress on their first job:

“Having the ability to not overcomplicate things [is invaluable].

“Start with a minimal viable product -- from a technical perspective also -- that will get you some influx. That’s far better than going for the ‘Rolls-Royce’ of solutions”.

4: Data Scientists are autonomous thinkers who excel on their initiative, whose job is to innovate

If you hire a Data Scientist, you have hired someone who can think for themselves, according to Burchell (“to do solutions in your company, to optimise a process, or do some sort of innovation based on data”).

And a client of a contractor data scientist can therefore absolutely expect the data scientist to “do” some innovation, like developing R&D, or spotting the potential for a new product.

'Whatever; you don't get to suggest the work -- it comes from the top'

But working at a “company that won’t have your back” once you spot that potential new product defeats the point of the company having a data scientist in the first place.

Burchell explained her assessment: “A company that says something like ‘Whatever; you don’t get to suggest the work -- it comes from the top.’ Well, [to such companies I’d simply say] ‘don’t hire data scientists.’ Because they’re going to get annoyed and leave.”

5: If treated like a kid, still don’t throw your toys out the pram -- or play with fire around a professional bridge

In some companies and not just companies seeking data science skills, the approach appears to be ‘we hire experts so we can treat them like small children,’ said WeAreDevelopers’ boss Rudi Bauer, at 47.05 of the online conversation.

But even if the hiring manager did treat you like a kid – and then you get to leave the role despite having found parts of it fulfilling, Greijn recommended: “Never burn a bridge where you really enjoyed yourself.”