Attended 10+ interviews for data science roles including #Faang , startups & mid size companies This is my experience so far Everywhere the interviews are like war , 1 person ( interviewer) knows some topic very well and he may be v good at it and asks questions related to it or only what he knows.. failing to answer it.. you are out of the game ..đ Especially for data related roles the general interview round include 1. Tech screening 2. Python / sql 3. ML / Statistics 4. Product sense 5. Take home / case study / presentation 6. Panel interview + behavior questions 7. Hiring manager 8. Sometimes with other team members or cross functional teams There is no limit to expect from 1 person they are hiring for. You need to know data analytics, dashboard development , statistics, machine learning, deep learning, computer vision, ETL, Data warehouse Spark , data engineering, model deployment, product management, AB testing presentation software engineering, Python, spark, R , sql , Java script, vba, master in excel, Tableau, power bi, Dax, GCP, AWS docker, mlops kubeflow. etc etc.. and people ask you have used Tensorflow do you know pytorch ? But their projects include dashboards, ETL tasks ..đ€·ââïž You need experience in marketing analytics, product analytics, finance analytics, ad marketing, sales analytics , e commerce, supply chain and what not.. Very Hard sql query which we will never use like that in the real world ( rarely if we are analyzing data from multiple data bases and integrate and cleanse them) but many can be solved easily in the real world looking at stackoverflow or mostly data engineering teams work on these tasks An interviewer picks a random ml algorithm and asks to implement it from scratch.. I mean why ? It's already available as a package in sikit learn..đ„ș As a data scientist we need to be problem solvers not a jackass memorizing all algorithms and forgetting it after interviews. It is a creative field and requires a lot of patience , thinking, working with people to understand business or domain knowledge or product knowledge and do experiments to come to a final answer or a model. Companies emphasize a lot on tools and technologies and I dont know if I'm feeling this way or how others are getting jobs and clearing these many interviews... I'm very overwhelmed with the expectations from the companies who are hiring a data scientist Edit :Many asked me which companies gave take home assessment , these are the ones 1. Greenhouse 2. Pocket games 3. Skillz ( most fucked up process recruiter said sorry other candidate accepted offer, after solving the assignment for 2 days and submitting it) 4. Tempus 5. Dow Jones 6 Move works 7 Tesla 8 Apple 9 Blue Cross blue shield Tc 100k #data #dataanalytics #datascience #interviews
Anyone who has this amount of diverse experience expectation is expecting too much from a given DS. You will NEVER have to do all of that in one go. That being said there are DS roles that specialize in each of those things. BUT they shouldn't be asking all of those for the same position - that would be like looking for a fucking unicorn
Lol I just failed FB DS on-site a couple months ago, and it's exactly what op described. Same with a few other on-sites I've done. I'm no way have the interviews been representative of the actual work I would do, which I can confirm because a close friend and former teammate of mine from a different company is currently a DS at FB. His exact words after I told him I didn't pass (he was pretty pumped to work together again) were "what did you do? tell them to fuck off?! You're way better at this stuff than I am, so if I can do it you'd crush it." It's actually quite ridiculous what they expect.
I feel like Iâve seen your post somewhere about that experience. But hey man, sometime itâs just luck and keep going !
I agree this is a common problem with DS interviews and makes me want to switch to SWE. SWE have it easy with all the interview prep resources out there and only expected to know LC and maybe system design
Only LC and system design
Yeah you only have to practice LC and Object Oriented Design and System Design and then when you join the meeting fully prepared they ask you to recite the entirety of Java's collection API, or explain all the tags used in SpringBoot, or open a Google dock and ask you to code a JavaScript Todo app and then design the backend REST API for it. Or ask some micro detail or gotchas of the programming language or framework they know about. Or force you to use an unfamiliar programming language to do the LC questions and if you make any syntax mistakes you are dinged points for it.
I Normally say I wonât do take home. Then they may come back saying we wanna move forward . Feel free to reject take home . Itâs a waste of time , not to mention itâs free consulting.
I always reject too I tell them too if thereâs too many interviews
Yes, I stopped when I was ask to do one for a company.
I had interviews like this before where interviewers ask questions of all different kinds of topics, very wide in breadth. that signals to me that they don't have a clue what to look for in the candidates, so the questions are all over the place. if you don't get the job, dont sweat it it's not worth your time.
Honestly OP the grass is not that much greener on the other side. Even SWE interviews are a black box, where you can be asked anything, it's not just LC and design. I've been giving a few interviews myself and it's amazing how many hoops you have to jump through to finally do shit that doesn't even come close to the complexity of the questions they ask. The really sad part about all of this is that it's not just for new grads, even if you get a decade of work experience doing these things, you will still have to go through this grind in the next interview.
The thing is DS has no proper role. I have given 15 interviews. Some of the companies asked me SQL in depth for screening test. Some had LC medium coding round for getting shortlisted. The project I did complex and challenging one didn't help me to get the job. They rejected me because I have never worked on NLP. Some rejected me because I have less knowledge on stats. Some wanted me to explain ML algorithm in depth with maths involved and how optimization is going on. It's like you need to know stats(practical one which I didn't have+ theory), ML algorithm (in greater depth , no YouTube videos covered in that much depth since my interviewer did master in stats he asked me in details), SQL, DSA(LC medium).... And so on. You can be asked anything đ€. @walmart: Wells Fargo's(Bangalore) interviewer didn't seem to be interested when I told him I have not worked on NLP but I do have some knowledge. Interview round concluded in 15 min only. I am planning to switch to SDE because my hard work is not reflecting in role and to salary. Plus the opportunities are less and proper roles are not there.
Same here buddy. I was interviewing with Ericsson India and that guy kept on asking me NLP questions even though I clearly said to him that I haven't worked on NLP. Then he rudely asked me which algorithm I am comfortable with as if haven't given answers to the questions he asked me LR, RNN, LSTM. I said currently I am working with Xgboost so I am pretty confident with that.. and he didn't asked any questions on thatđ€·. Then why u asked me which algo I am comfortable with. Bcz of these jack*** people we feel down and think we won't be able to crack DS interviews.
Data scientist is like the full stack developer term way back. The role is completely based on the company. I'll stick with my business intelligence moniker.
What's the minimum I need to know to get into BI?
SQL, data visualization tool (Tableau, Power BI, SSRS, QlikView, etc ), data/business analysis. Good people skills like experience in working with c-suites or user enablement/education. Python and/or R are great to have.
I agree. Total lack of any standard. You donât even have a clue what youâre expected to prepare
Hey OP, our DS team is hiring and Iâd love to refer you. Iâd like to think that we have a smooth interview process! Shoot me a DM.
100% agreed.
Why not apply for software engineer positions?
Maybe because he is already pigeonholed into DS and can't get an SDE interview at a FAANG?
Two different skillets, DS usually are not into LC and its easier to pick the other related skills OP mentioned in the post than the dreaded LC