How To Secretly Interview the Interviewer

[Music] hey everyone got a little change of scenery for you today um this is going to be a unstructured video no real script here today but i do want to talk about something that's been on my mind lately which is data science interviews now when you think about a data science interview or any interview really typically you're thinking about it as i'm the candidate i'm going to go meet somebody at the company and they're going to evaluate me and my abilities and this is definitely true to an extent but i think there's a flip side to this that people don't think enough about but you definitely should be thinking about this i would argue even more so which is that just as the company is trying to evaluate something about you you should also be using the interview to evaluate the company now this sounds a little weird and it also sounds difficult if you think about a interview let's say last one hour typically 45 or 50 minutes of that hour are you getting asked questions by the interviewer explain this to me code this so and so so it's only that last 10 or 15 minutes where you explicitly get to ask questions to the interviewer to get what i like to call explicit signals about the company i call these explicit signals because it's explicitly you asking a question and hopefully getting a good response from the interviewer now we'll talk at the end of this video about good questions to ask in that last 10 or 15 minutes but because most of the interview 45 or 50 minutes of an hour are you getting asked questions i think it's equally or actually more important for you to get really good at reading what i like to call implicit signals from the interviewer and i call these implicit signals because they're not any information you gather from asking questions instead it's information you're gathering in the background or in the context of the interview while it's happening and this can be a difficult skill but i think it's something that you can learn over time as you do more and more interviews and it's also something that's extremely important to be able to gather to know if this company that i'm interviewing at is somewhere i would even be happy working where my skills would be valued and if it's somewhere i want to spend the next near future of my life and so let's talk about implicit signal number one and this one comes during the part of the interview where they ask you about more theoretical data science concepts explain this concept to me or explain the math behind this stuff like that you know what let's do it this way let me give you what i consider a bad set of questions that you might get during this section and talk about why i think it's a completely insane set of questions that you could get asked this would be a bad set of questions let's say the first question is explain logistic regression to me let's say you answer that the next question is explain decision trees to me you answer this one and let's say the final question in this section is explain pca or principal components analysis to me on the surface these seem like great questions i mean each of them is geared towards some area of data science and maybe you should need to know them for the job think about it one level deeper and and let me lead with this let me say that any data science job no matter what field or technologies or tools you're using all have this one thing in common which is that data scientists are at the end of the day people who are solving complex problems and solving a problem is inherently something that requires step-by-step thinking kind of going from something abstract to getting something concrete at the end of the day think about this set of questions each of them individually may have something to do with the field of data science but the key insight is that these questions have nothing to do with each other they don't build on the last one they don't tell you anything about the candidate's problem-solving ability and so in the moment i'm not saying you should cut the interview short and say this is a bad set of questions how dare you i think you should answer these questions to the best of your ability but after the interview i do think it's worth critically thinking about what did the interviewer actually gain learn about my problem-solving ability based on these questions and the answer is probably nothing these are questions that i'd probably get on the top google search result for list of data science interview questions it doesn't seem like there was a lot of thought put behind the sequence of questions and so let me offer maybe an off-the-cuff response for a better sequence of questions that you might get during this part of the interview the first question may still be explain logistic regression to me at a high level but the next question should build on that it may be something like explain the math behind logistic regression explain the loss function what is it trying to do why does it work so this is kind of getting into the mechanics of the algorithm and maybe the final question is more geared towards how do we actually translate this mathematical thing into some real world decisions so we might ask something like um how do we take the probabilities that come out of logistic regression and use that to make some kind of business decision so do you see what i mean like this sequence of questions is just infinitely better than the last sequence of questions because each one is building on the last one you are evaluating the candidates problem solving ability how do they think about these algorithms how do they how do they approach the relationship between the mathy side of data science and the more practical decision making business side of data science this is not things you're going to learn from that initial series of questions and so that's why i consider this implicit signal number one now let's talk about the other big implicit signal which is something you would see during the coding section of the interview so as you know data scientists still do need to code and so it's very likely you'll get a coding section of the interview and they'll probably just ask you something like uh here's a problem that i want solved make your best attempt at coding up a solution to this problem now even if you have prepared for these coding interviews beforehand even if you've done tons of practice it can always it's normal for it to happen that you just get completely stuck at the beginning now as counter-intuitive as this might sound while you're kind of sitting there stressed about how to solve this problem the the other thing you should be thinking about is how is the interviewer reacting to the situation of me being stuck in my experience it can go one of two ways and here's the bad way is that you get stuck and i think the maximum time would be around five minutes before you make this call the reason is because typically these coding exercises are 35 45 minutes so five minutes is a rather long time to get stuck if the interviewer has not said a word has not stepped in has not done anything in five minutes or even worse if you've asked them for help and they've explicitly said no i'm not going to help you i think that is a bad outcome and that's an implicit signal that maybe i don't want to work at this place and let me tell you why in any real data science job you're not going to be working alone hopefully you will have a team that's going to be working with you they'll be helping you with stuff you'll be helping them with stuff and so we know that this role should be team based should be interactive now of course the interviewer should give you some time to think about this problem that is normal and expected and people need time to think through solutions but at the same time they should step in and offer that little bit of help that you would get in the role anyways to kind of get you unstuck from the situation i think that if they're just willing to let you be stuck for a long amount of time and risk not learning about what you could code up if you did get past this one particular hurdle that doesn't send very good signals to me so i do think that the other implicit signal to watch out for is that if you get completely stuck during the coding part of the problem do they step in give you a gentle nudge in the right direction get the ball rolling and help you kind of solve the problem or does it seem like you're against the interviewer where it's like you versus me you never want the interview to feel like too much of a you versus me situation i think it's better if it feels kind of like here's a challenge that i'm proposing you take your best attempt at it but i will help you get the ball rolling if things aren't going well so that is the other implicit signal to watch out for and finally as i mentioned in the beginning of this video let's talk about the best way to gather explicit signals at the end of the interview so at the end of the interview they'll give you 10 or 15 minutes hey do you have any questions for me sort of thing this is an extremely personal decision it depends on what you want out of this job for example let me just give you a couple examples if you're somebody who really values teamwork let's say that you would not thrive in a setting where it's mostly individual work ask that question and and what i like to do also is ask it in a way that makes the person think about where the deficiencies might be in that metric for example instead of asking is the company collaborative you might ask what are the biggest challenges in collaboration between people at the company and that kind of gives you a better lens into where the deficiencies might be and how they're going about addressing it so that's just one thing uh maybe instead of collaboration you are more of an individual person where you care about having enough time to kind of heads down do your work and kind of code up a really cool solution to a problem then go ahead and ask that question instead you might ask something like how do you ensure that people have time to get heads-down work done or have time to think about problems deeply on their own if needed so whatever the thing is that you care about ask that question instead of asking some canned question you found on google that you think is going to impress the interviewer definitely use those two or three or four questions the small amount of explicit signals you can gather to understand things that you personally care about not what somebody else cares about and so that i think should wrap up the video uh again super unstructured so i hope the points i made were reasonably clear but the main point that i wanted to make is just that an interview goes both ways the way we typically think about it is the company interviewing you judging something about you but it should be even more so about what am i learning about this company through the interview and part of that is implicitly and part of that is explicitly but at the end of the day after your interview you should take some time and reflect on is this a place i would want to work based on these signals that i've gathered and hopefully that makes a little bit of sense so anyways if you like this format please let me know if you hate this format please let me know and like and subscribe i'll see you next time

Channel: ritvikmath

Views: 954

Keywords:

Id: rEmOO7vHb5I

Length: 10min 7sec (607 seconds)

Published: Mon Dec 20 2021