Alan Turing a pioneering recruiter? You wouldn’t normally think of the father of modern computer science in those terms. It’s more natural to associate him with conversations on Artificial Intelligence – after all, he shaped our understanding of how computers solve problems, and how their ability differs from that of human intellect.

Where does recruiting fit in?

I loved the movie The Imitation Game, where Turing turns around Britain’s wartime effort to crack Nazi Germany’s Enigma encryption scheme. If you haven’t watched it yet, you absolutely need to. Here’s what caught my attention in the movie – Turing was clearly a man of exceptional ability, yet he didn’t pretend that he could win the game by himself. He knew what he was up against, and set out to assemble a winning team. Part of his success involved using novel approaches for finding talent.

What lessons from his story can we use when recruiting data scientists?

Cast a wide net

“Sometimes it is the people no one imagines anything of who do the things that no one can imagine” ― Alan Turing

Turing started out by casting a very wide net. He published challenging crossword puzzles in the London Daily Telegraph and let problem solvers self select – his approach was unique because it emphasized exceptional problem solving ability and speed, rather than professional background. Among the codebreakers he recruited was the incredibly talented Joan Clarke. As a woman, she would have been overlooked on account of prevailing biases of the time. Similar attitudes are still common today, and that makes objective screening methods extremely important.

In my previous blog post, I analyzed LinkedIn data to illustrate typical career paths that lead to data science. What you can do is reach out to promising candidates in such fields who aren’t yet marketing themselves as data scientists. A good signal would be if they mention on their profiles any data science classes they’ve taken – while taking a class doesn’t demonstrate job effectiveness, it can suggest interest and motivation. And a commitment to always be learning.

Design take home tests relevant to your business problem

Data science teams typically use take home tests as the next level of filter – bite sized problems that allow candidates to showcase their skills, while also letting them use their favorite reference material, much like they would during a regular day on the job. I like how that takes the pressure off potentially nervous candidates, and allows them to focus on problem solving instead.

Turing’s “take home test” was the crossword puzzle. Why crossword puzzles? Because solving crossword puzzles requires the same style of thinking that you’d need for reverse engineering cryptographic codes and ciphers – a knack for getting into the mind of the puzzle creator, and anticipating ways in which he might be trying to mislead you. Crossword puzzles were a practical way to screen for the right skills without giving away what the British government was up to.

What does this mean for you?

You need to ensure that your take home tests actually reflect your day to day challenges. Don’t waste everyone’s time assigning problems straight out of a textbook.

For example, if your main challenge at present is collecting data and ensuring quality, test for that. How well can candidates use data APIs and web scrapers to collect data that’s not readily available? How do they profile collected data and handle missing or outlier values? Do they make an effort to improve data quality by combining other known high quality sources? How do they account for quality issues when gauging the correctness of insights they uncover? Provide hints to guide them, but then give them the space to connect the dots.

What’s in it for the candidate?

Recruiting is a serious time commitment, both for you as well as for your candidate. The take home test is your first line of defense in identifying bad fits early on. Likewise, what’s on your take home test allows perceptive candidates to spot data science teams who don’t know what they’re doing. If you’re not already a powerhouse brand like Facebook, Uber or Airbnb, put yourself in your ideal candidate’s shoes and ask yourself – “Why would I want to interview here? Should I be spending my free time on this take home test? What will I get out of it if things don’t work out?”.

Do what Alan Turing did. Invest the time in putting together a good take home test. Have your team members take the test and iron out the kinks. Candidates should feel that you’ve put an honest effort into it, and that they can learn something unique about your business and data science. Appeal to their curiosity and thirst for a worthy challenge. Make them want to come interview with you.

Happy recruiting!