How to design an Interview process for Data Scientists
Data Science Recruitment Guide part 2
How to design an Interview process for Data Science Candidates
What are you trying to achieve in hiring this Data Scientist?
What do you want this candidate to achieve in the first 6 -12 months on the Job?
They seem like simple questions and they are, but everything in your interview process flows from them.
Points to be aware of before you start designing the process – Bias and how to manage it is a topic you should be aware of,
- To attract as many people from as wide a background as possible is good for your business, therefore being aware of what might put candidates off is important.
- There are various types of Bias Information, Selection & Confounding
- Most relevant are probably age, disability, ethnicity & gender these should be relatively obvious, but Likeability is often overlooked you are not just looking for someone you get on with and makes you feel comfortable – assessing skills and critical thinking are the key.
- Just focusing on easily definable but often very shallow criteria – that often define many job specs (5 years’ experience with C++ etc) these are basic frameworks to build the job spec around not assessment of skills.
Design an interview process by building a framework of key metrics and how to assess them at the beginning of the process. Amazon famously has its Bar raiser program which includes measuring key metrics and demands a specific level of performance above a set benchmark before a candidate can be hired – this includes asking random and awkward questions, to see how people react to unusual circumstances under pressure.
Copying another company is not advised as you want something bespoke to your values that shows a clear reasoned & consistent strategy to the hiring process.
Other aspects to consider are – The overall time the process takes
- The longer the process takes the more likely you will have candidates drop out. Ideally, you want to maintain regular contact with the candidate meaning relatively short gaps between each stage as well as having a clear process laid out to candidates.
- Adding extra stages unexpectedly to the process either because you have not gained the knowledge you needed from earlier stages or because you need additional approval before making the hire, all denote a failure of some part of the process.
People are always busy and scheduling can be tricky
- key people will be needed for an interview panel as well as the candidate, plan for this.
- Make sure each person gives it the priority it deserves.
- You will pay for this later if there are undue delays.
- Or worse people jumping in & out of the meeting will make assessing of results patchy.
- The candidate experience might also be damaged making them less likely to choose your business.
Never forget interviews are a two-way street and candidates will be assessing you as well, just as you expect them to make an effort to prepare and impress you, they should expect the same courtesy.
However, shit happens so in exceptional circumstance the benefit of the doubt should be given.