My favorite technical interview question by Philippe Barbe 22 Sep 2021

I am sometimes asked what questions I ask in technical interviews. Over the years I have seen many interesting questions that reveal skills and knowledge.

What is the question for?

An interview question should be evaluated by comparing what the interviewer is seeking to know about the candidate, and how much relevant information an answer will likely deliver. Therefore we first need to consider what the goal of a question is.

If I want to have an idea on how well a candidate knows a computer language, I may ask questions about the intricate parts of that language.

If I want to know how quickly the candidate codes, I may ask a coding question.

When I hire a Data Scientist, I ask how well the candidate knows Python. How quickly the candidate can write code is also something I care about.

But these are not the things I care about the most. Therefore, this type of question is not my favorite.

Questions on specific techniques all achieve a similar result: they tell me if the candidate knows these techniques. There are three problems with this type of question:

• They tend to reduce the job to knowing a few topics. I find this overly restrictive, since the nature of R&D is that you do not really know beforehand what the most useful tools may be. When it comes to Data Science, I am seeking versatility, structured thinking, and creativity. I am seeking colleagues who can imagine solutions, more than team members who can reproduce existing patterns.
• I can only ask questions about topics that I know. I may have learned a fair amount about a lot of topics over the years but still what I know is a tiny fraction of what I don’t. To build a balanced team that can field any problem, I do not want copies of myself. To the contrary, I would rather get people who complement each other. Being limited to asking only about topics I know doesn’t achieve that balance.
• If the candidate does not know the topic, then what? What does it say other than the candidate does not know that specific thing? Of course, if the knowledge of that topic is foundational then that candidate won’t be hired. If it’s not then it’s simply an unfortunate circumstance which speaks more about the places where the candidate has been student and employee than about their abilities.

Certainly there is core knowledge that I need team members to know about so I have to ask questions about it. This is why I may ask a coding question, to make sure that the candidate knows the language they claim to know.

This is also why I may ask questions around various topics. They give an idea of how well-rounded a candidate is. There is value in knowing that.

Nevertheless, you may not know something I deem important for the job and still be a fine candidate worth hiring who will, in fact, be a great team member.

What am I looking for?

What I really care about is that the candidate can learn… and how quickly, how well and how deeply?

I could ask “Can you learn?” but I already know the answer. Data Science candidates for an R&D team have a Master’s degree, and likely a PhD. They’ve learned all sorts of things including the art of résumé writing well enough to pass a recruiter! Sure they can learn!

But, is the candidate a fast learner? That is a subjective judgment call that is very relative. I have yet to meet candidates who tell me upfront that they are slow learners! Besides, one can be fast at learning something and slow at learning other things.

That leaves “How well can you learn?” Presumably, people perform the best doing what they know best. I know the candidate can learn and assuming I giving them the time to learn on the job, how well they do their job on my R&D team will likely relate to how well they can master a topic.

Therefore, what I really want to know is how well they know what they know the best.

Presumably, this tells me something about the best performance level they can achieve. Unfortunately, there is no point asking “How well do you know what you know the best, on a scale 0 to 10?” The answer will always be 10 since it is what they know the best. And if I just ask them to quantify their best knowledge, I would not even know what it is they are best at but only the level!

Conclusion

The reality is that I do not have a favorite question for a technical interview! What I do have is a favorite, and very clear, command with which to start a technical interview:

Pick a technical topic that you know really well, and tell me about it.

What happens then? Read my next article!