How to Stand Out in Your Data Scientist Interview
TL;DR
The best interviews are when you and the interviewer have a conversation, not an FBI interrogation. Structure your answers like the examples below to manipulate the dynamics of an interview, so that you make the interviewer feel like they just had an intellectual chat with a colleague.
Why you need to read this article
There are already a million articles out there that provide information on Data Science interviews, from what to expect to what technical questions to prepare for.
They let you know their typical structure, what technical questions to expect, and all the keywords and concepts that you need to put into your response.
However, they rarely talk about ‘how‘ to deliver your answers.
And I believe the delivery is key.
Having been both interviewer and interviewee, time and time again have I witnessed that it is not all about your knowledge.
More crucially, you need to know how to deliver your knowledge effectively. Interviews are human-to-human interactions, inevitably influenced by human biases.
A candidate who gives a monotonous drone containing a word-perfect response is going to be disadvantaged compared to someone who gives an answer identical in content, but in a structured, eloquent and interactive manner.
In addition, the best interviews should feel like a conversation rather than an FBI interrogation. If you are doing most of the talking, with little to no engagement from the interviewer (except for when they ask the questions), then most likely the interview is not going well.
So, let's dive into how to manipulate the dynamics of an interview and how to deliver your responses to maximise your chances of landing the job.
The key concept you must understand – Conversation
A common phrase interviewers say before starting an interview is:
"this interview is meant to be a two-way conversation, … it is an opportunity for yourself to get to know the role and company as much as it is for us to get to know you…"
However, it does not change the fact that a significant majority of information exchange is from you to the interviewer.
But, even if 90% of the information exchange comes from you, you cannot allow yourself to be speaking for 90% of the time.
The key concept is that you need to purposefully structure your answers such that there is room for interviewers to engage with your response as though it was a conversation.
Take the following example:
Let's say you have 10 years of experience as a Data Scientist in , having worked in 5 different companies (A, B, C, D, E) for two years each. You are a Senior Data Scientist, applying for a Lead Data Scientist role.
Example Bad Response
The interviewer opens up the interview with an ice-breaker question:
Q. Could you give me an introduction of yourself?
You respond with:
A. Yes sure. So I've worked in company A as a Data scientist, working in the industry and I worked on a project for , which involved building a model for … .
I used features because was relevant for .
Then I moved to company B, where I worked on using models … .
…
Then when I was at company E, I did …., using model using approach and for hyperparameter tuning…
You might be reading this and thinking "yeah… that's so obviously bad, who actually does that".
Or perhaps you're not sure what is wrong.
The fact is:
- The interviewer already knows your experience from your CV and your cover letter. They have come prepared with questions to ask you on particular points they are interested in. These points are what makes you interesting to the interviewer.
- Meanwhile, the above response is likely to be lengthy whilst not being able to hit those points of interest, or it would take a very long time to do so. Given that you have no visibility on what the interviewer wants to hear, you are going through every single detail of your career that does nothing to impress them.
As a result, two negative outcomes can occur:
- You either talk for too long and the interviewer loses focus and/or cuts you off as they need to ask you the questions they need to ask in the allotted time.
- Or they let you talk, and the opportunity to talk about the things they are really interested in goes out the window.
In this situation, the key is to get the interviewer to ask you what they want to know from your experience.
So how do you do this?
Take the next response as an example.
Example – Good Response
Q. Could you give me an introduction of yourself?
A. Yes sure. I have been working in the industry over the last 10 years, having worked on projects such as
, and . These projects have involved the use of and , and the successes of those projects have led to and . If you have any points that you wish me to elaborate on, or if you had particular questions about what I have said so far, please let me know and I will be happy to expand on them.
This response is divided into two sections:
- A very high level overview of what you have done.
This is not only a summary of your work, but it should be the things you are most confident in speaking about, and the things you anticipate to be the points that the interviewer may be interested in.
2. Purposefully let the interviewer choose the next topic of discussion
This is the most important part of any response. You now need to let them choose which road to take, and this is what changes an interview from being a one-way information delivery into a proper conversation.
- It keeps them engaged.
- It allows them to dive into the things they want to hear more about.
- It gives you the opportunity to shine and appeal your best work that is most relevant for the role.
Interviewer response: Yeah, I wanted to ask you about
actually. Could you explain to me how you…
With the first ice-breaker question, you've managed to get their interest, and nicely set yourself up to highlight your best strengths.
Apply the same principle to technical questions / case studies
The same principle can be applied to technical and case study questions.
Systematic ways of structuring your answers to such questions has been dealt with in many other articles. Here, I quote the following article as I mostly agree with their approach and it is something I have also been using:
In it, they outline the ASPER framework:
1. Ask. Ask questions to uncover details that were kept hidden by the interviewer. Specifically, you want to answer the following questions: "what are the product requirements and evaluation metrics?", "what data do I have access to?", "how much time and computational resources do I have to run experiments?".
2. Suppose. Make justified assumptions to simplify the problem. Examples of assumptions are: "we are in small data regime", "events are independent", "the statistical significance level is 5%", "the data distribution won't change over time", "we have three weeks", etc.
3. Plan. Break down the problem into tasks. A common task sequence in the data science case study interview is: (i) data engineering, (ii) modeling, and (iii) business analysis.
4. Execute. Announce your plan, and tackle the tasks one by one. In this step, the interviewer might ask you to write code or explain the maths behind your proposed method.
5. Recap. At the end of the interview, summarize your answer and mention the tools and frameworks you would use to perform the work. It is also a good time to express your ideas on how the problem can be extended.
The Ask, Suppose and Plan sections can be done without interviewer engagement.
However, I don't think it wise to jump straight from Plan to Execute, for the following reasons:
- There may be particular points in your plan that the interviewer is interested in (but you don't have any clue what they are).
- Your interviewer might not agree with some of the steps, but again, you have no clue if this is the case.
Therefore, the step between Plan and Execute is an ideal opportunity to put the ball back into the interviewers court, and so we add in an additional step into ASPER, called:
3.5 Ask for Feedback. Announce your plan, and ask the interviewers if they have any intermediate questions to ask.
Again, this step keeps the interviewer engaged and works to change the dynamics of the interview from being a one-way information regurgitation into an intellectual conversation.
If the interviewer has no questions, you're in the clear – you may proceed with your answer.
If the interviewer does raise some questions, then listen carefully – the interviewer is trying to steer you into the right direction and it is up to you to catch onto this quickly and to adjust your response accordingly.
Summary
In order to shine in an interview, you need to know what grabs the interviewer's attention and what makes you interesting to them. To do this, you need to specially structure your responses to naturally squeeze this information out of them.
Be it an ice-breaker, technical or case study question, use the techniques we have covered above to make yourself stand out above the rest.
The best interviews are when you and the interviewer have a conversation, not an FBI interrogation. Structure your answers like the examples above to manipulate the dynamics of an interview, so that you make the interviewer feel like they just had an intellectual chat with a colleague.
