4 key traits to look for when hiring a Data Scientist

towards-data-science

This post was originally published by Richard Farnworth at Towards Data Science

Plus sample questions to use in an interview

Image for post

Photo by Glenn Carstens-Peters on Unsplash

Finding good Data Scientists can be a tricky task. Googling “Data Science Skills Gap” shows that in many countries around the world, companies are struggling to find suitable candidates for their ever growing Data Science needs.

A bad hiring choice can be very expensive, and as a result, it’s important to be able to vet candidates effectively, to make sure they are a good fit for the position and are going to be effective in introducing/expanding Data Science at your company.

Rigour

Along with rigour, goes curiosity. A good Data Scientist will ask many questions of the data they’re working with, and before long should understand its properties, limitations, quirks and interesting internal relationships. Data, after all, describe something happening in the real world, and taking the initiative to dive into the data and discover its secrets opens up new opportunities beyond the original brief.

Sample questions:

  • “Give us an example of a time when a data quality issue impacted your work, how you discovered it and how you dealt with it.”
  • Set a practical question using a publicly available dataset that is relevant to your industry. A good sign to look for is if the candidate asks many questions to confirm their understanding of the data, before they jump into offering insights.

Resilience

But often, the most difficult thing to deal with is uncertainty. When you begin a modelling or analytics project, there’s no guarantee that you’ll be able to find anything useful. No matter how much time and effort you put into the endeavour, the data you have might just be all noise and no signal.

When nothing seems to be working, it can be frustrating and disheartening, and so resilience is an important trait to look for. It’s critical to have the ability to keep pushing and trying new things without giving up, or worse, compromising on your professional integrity. But it’s also important to maintain a clear commercial head and know when to wrap things up once the law of diminishing returns renders any further effort futile.

Sample questions:

  • “Tell me about a time when a piece of work you were excited about didn’t go to plan.”
  • “How do you decide when a piece of work is ready to be delivered?”

Empathy

Practicing empathy is therefore a core skill of Data Science. Understanding the particular experiences and desires of everyone involved and communicating effectively are critical to a project’s success and should not be deprioritised in favour of technical skills.

Sample questions:

  • “Tell me about the people involved in a recent project, and what their differing expectations were?”

Technical Skills

The ideal candidate will have 10+ years of SQL, Python, SAS, TensorFlow, Keras, PyTorch, Tableau, PowerBI, Qlikview, yadda yadda yadda, all within an Enterprise-Level Finance environment delivering to C-Level executives”.

The technological landscape of Data Science is changing all the time. As such, the most important ability is being able to quickly pick up new technical skills and fit them into one’s repertoire.

Experience in a specific technology may be particularly important in your company, so it’s fine to state that in the job description. But try to keep things loose, so as not to exclude technically talented candidates, who may not strictly match that experience, but will pick up the required abilities quickly.

Sample questions

  • Don’t rely on self-proclaimed experience or LinkedIn endorsements. Not all “experience” is created equal! A practical question, working through a case study from your industry will tell you much more about how a person’s thought process works and how they apply their technical skills to a problem. For example, I’ve interviewed several candidates in the past who claimed proficiency in SQL, but struggled to apply the concept of a JOIN when presented with a handful of tables and a question about the data.
  • Sites like HackerRank offer testing environments in many programming languages, which can be useful as verification in the early stages of thinning down candidates. Test each candidate on their self-described strengths, rather than just the skills listed in the job description.
Spread the word

This post was originally published by Richard Farnworth at Towards Data Science

Related posts