3

I'll soon be completing studies. So, I have been applying for jobs. Generally, I'm able to cross the initial pre-screen assessment test by HR personnel, in which they typically ask the tell me about yourself? question.

For the technical round, the company will give a dataset and will either ask a specific problem to solve or will ask to explore the dataset and then conjure up a problem to solve it. Generally, the dataset size is quite small and easy to process.

Recently, I interviewed in a company for the role of a data scientist. I crossed the pre-screen HR round. For the technical round, they gave me a real dataset whose size was over 200MB! They said,

The following challenge asks you to work with a data set of loan repayment. It is intentionally meant to be open-ended. The point is not to arrive at a predetermined answer or search for the lowest possible standard error. Rather, the hope is that it will force you to ask relevant questions about the data, do some preliminary exploration, perform the necessary manipulations or aggregations, generate visualizations, and reach conclusions or insights. The most important thing to remember is that we are evaluating your thought process and ideas! The more you explain your thinking, in a clear and succinct manner, the better. If you get stuck, describe what additional information or data you might look to collect, and trying a different idea is highly encouraged.

I provided a comprehensive 50-page report in which I formulated a problem, using the given dataset and explained in great depths on how I solved it. I also explained the given data anomalies, how to eradicate them in future. In short, I think the report was so rich in content that it can be accepted for publication!

The company rejected my job application and did not provide any valid reason for rejection (Note: I wrote back to the company post the application rejection seeking feedback and they have since not replied). Now, I understand they are not bound to provide a rejection reason, but what really hurts and baffles me are the following questions;

  1. In future how to deal with such open-ended technical round interview questions? Should I provide a detailed report or not?

  2. Is it a red flag to be given huge datasets and open-ended technical problems to solve? How to determine the company is not exploiting a prospective candidate in terms of acquiring a free end-to-end solution?

  3. For future technical round interview, is it appropriate to ask something like, Will you provide a feedback post completing the technical exercise?. If the company is reluctant or refuses to provide feedback, what should I do?

  • 2
    Sorry if I'm going to come across as blunt, but what makes you think 50 pages is succinct? – lucasgcb Jul 11 at 7:56
  • 1
    @lucasgcb you've raised a debatable point. Let's not get into it, for it will be a deviation to the asked question. Rather, I hope you might consider adding something meaningful to this discussion! – mnm Jul 11 at 8:06
  • 2
    I mean, they set up some requirements and a long report goes against that. In your question, you still write it as "comprehensive" despite so, which shows you haven't noticed it as a problem. I'm trying to take into consideration what is your understanding of what you did before formulating an answer. – lucasgcb Jul 11 at 8:12
  • I would rather ask how to deal with the few that actually do give feeback! Most companies won't talk to you again after rejection except for a generic message. First, because it takes time for them that they don't want to spend on a candidate that won't work for them. Also, every reason they may give is a potential lawsuit – David Jul 11 at 13:27
  • I won't post this as an answer because it doesn't answer your question directly but if I were assessing candidates I don't think I'd have time to read too many 50 page reports. Maybe it was a case of TLDR. – Old Nick Jul 11 at 15:44
5

In future how to deal with such open-ended technical round interview questions? Should I provide a detailed report or not?

The way to find out what the interviewers expect from you is to ask them.

A 50 page report that could potentially be published sounds like far too much to me (although I'm not a data scientist). If it was real data then you've just done a lot of work for free, and if it was just an example data set then someone would have to spend a lot of time reading and evaluating your report, time which they most likely don't have.

Is it a red flag to be given huge datasets and open-ended technical problems to solve? How to determine the company is not exploiting a prospective candidate in terms of acquiring a free end-to-end solution?

Personally I think it's a good sign - your potential co-workers will likely have been tested as well, which improves your chances of working with a competent team in the event that they hire you.

It's common to find people who object to large technical tests though, and who aren't prepared to do them.

In the future I'd suggest you time-box the work you do on tests a lot better. That is, don't work for hours and hours until you have a solution that you feel is as the same quality as the best you would do if you were employed. Instead decide on a reasonable amount of time, and do the best you can within that time. The goal is to show that you're technically capable, not to deliver a complete production-ready solution.

For future technical round interview, is it appropriate to ask something like, Will you provide a feedback post completing the technical exercise?. If the company is reluctant or refuses to provide feedback, what should I do?

You can ask, but don't expect them to say yes.

If they reject you without feedback then accept it and move on to the next application. Rejections without feedback are a frustrating but normal part of job searches.

  • I did not understand the phrase, time-box the work you do on tests a lot better. Please explain it. – mnm Jul 11 at 5:50
  • @mnm I've edited the post, hope that's clearer now – Player One Jul 11 at 5:54
  • thank you it is clear now. – mnm Jul 11 at 5:56
6

A company will not provide feedback because it has the potential to get them into lawsuits and it has literally no benefits for them. A risk for zero gain. So they won't do it.

In the future, you might ask how extensive this report has to be. 50 pages seems long to me, especially considering that they need to evaluate many candidates. But maybe it's what they wanted and somebody prepared a better 50 page report. We cannot know, that's why you need to ask before you do all the work.

On the bright side, whoever they hired instead, that person will not compete against you on your next job application/interview.

2

We should start by dismissing attempts to solicit feedback after receiving a rejection. It should be clear from other answers and from the many other similar questions on Workplace that you likely won't get feedback.

Let's instead get to the crux of the matter. In the hiring process, it's not always the answer that counts - sometimes, it's the process of getting there. The interviewers gave you a big, explicit hint about this. I've added emphasis on the important parts:

The point is not to arrive at a predetermined answer or search for the lowest possible standard error. Rather, the hope is that it will force you to ask relevant questions about the data, do some preliminary exploration, perform the necessary manipulations or aggregations, generate visualizations, and reach conclusions or insights. The most important thing to remember is that we are evaluating your thought process and ideas!

It seems very clear that they did not want you to simply run off with the data, do a giant project, and give it back to them. They were more interested in your process than in your result. You basically didn't involve them in your process - you just handed them a result (which they basically said they didn't care about.)

When in interviews as a candidate, I remind myself: think before you speak. This applies to technical tests, as well. The point is, understand the interviewer's frame of reference before responding. It's almost always better to ask clarifying questions rather than simply responding with something that may or may not be relevant. When in interviews as a hiring manager, it's shocking how often candidates will immediately start speaking (or immediately start working on a technical test) without pausing to make sure they actually understand what they're supposed to be answering or doing.

You started your question by stating,

I'll soon be completing studies.

This is relevant because you're showing a tendency that seems common in new graduates. When you're in school, you're often given well contained and well defined problems, and your job is to go off, solve them, and provide an answer. In terms of tests, this is practically beaten into you - no one would dare stand up in the middle of an exam and ask the professor to explain a question to them! This works well in an academic setting, where the goal is to evaluate your results (so the problems have to be self contained and well defined). However, this can cause problems, because it trains you to immediately begin work when given an assignment, rather than pausing to validate the assignment and understand the actual business context. Many new graduates approach the interview process just like they approach an exam. For most employers, that's not really the intent - interviews are meant to be two way streets, where both parties interact and ask questions.

In the real world, it's typical that the people giving you problems (i.e. the business units you're supporting with your data science work) don't really quite know what they're asking for, and none of them can hand you a nicely packaged homework assignment. As such, an important part of the data science process in many organizations is not the data science work itself, but rather the interpretation of problems and refinement of an initial request into something that's actually solve-able and valuable. Based on how the interviewers phrased your assignment, it seems clear to me that what they were intending was to test how well you do at that part of the job, not just how good your data science skills are.

In my current role, I oversee a team of data scientists, analysts, and developers in a financial institution. It sounds similar to the type of position you're interviewing for. While I wouldn't quite give out a technical test the way they did, and I can't claim to speak on behalf of the people who rejected you, I can "get inside their head" in terms of their thought process. The next time you're asked something in an interview, or given a technical assignment that's worded like that, consider taking the following steps:

  • Stop and think about the context before you respond or work on the test.
  • Consider the wording the interviewer is using, and how that might give you hints on what they are (or are not) looking for.
  • If you're unsure, ask clarifying questions.
  • If appropriate, explain your approach before you do any work.
  • Consider the size/scope of your response as one of the things to potentially clarify - if they're getting 3 page summaries from other candidates, your 50 page analysis will seem out of place.
  • 1
    many many thanks for taking the time to elaborate on each and every point. It all makes perfect sense. I have to agree I simply jumped into finding a solution rather than making a conscious effort to think! A lesson well learnt. And heartfelt thanks to this community who took the time out to explain in detail what I was doing wrong and what can be improved. – mnm Jul 11 at 14:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.