Linkedin ML Internship

Interview process/preparation
#1

Hi everyone,

So I recently applied to the LinkedIn AI Engineer position and I got notified today that they wanted to move forward with an interview. They said that in the technical interview, I can expect to be assessed on Basic probability and statistics, Machine Learning fundamentals, such as linear regression, overfitting/regularization, clustering methods, ranking methods, natural language processing and Mathematical modeling . I’m feeling kind of nervous because I’ve never done an interview like this before. I was wondering if anyone who’s went through the process had any preparation tips or resources.

Thanks

#2

Hi there. First of all, congrats and try not to be too nervous - just prepare as much as you can! I assume your major/background is not in stats?

In terms of preparing for the technical interview, here are the major parts to review and prepare for:

  1. ML coding: practice writing code from scratch for methods like k-means, KNN, Gaussian Naive Bayes, etc. How would you do kfold cross validation?
  2. ML fundamental concepts: I would review the basics obviously as you mentioned but also deep learning (check out deep learning specialization course on Coursera), SVM, MLE, logistic regression. It’ll be good to know techniques to improve or tune your network such as dropout, batch normalization etc. I would say logistic regression, linear regression (including Ridge and Lasso), clustering and neural networks are most important. Know regression VERY well, there are a lot of questions that can be asked here. Also know the metrics in which to measure the performance of these models such as precision/recall, mean squared error, etc. They also like to ask about things related to Bayes Statistics/Bayesian interpretation. You wanna be able to explain the pros and cons of different learning methods and which one you would use given different scenarios or datasets.
  3. ML System design：you can check out educative.IO’s system design course, especially the par on recommendation systems.
  4. Probability and other math/statistics concepts: make sure you know things like random variables, expectation & variance, most commonly used distributions (normal, exponential, binomial etc.), conditional expectation, Bayes Theorem. You can also check out https://datascienceprep.com or any intro to probability/stats course on MIT Courseware.

Useful books to check out:

  • An Introduction to Statistical Learning by James et al.
  • Elements of Statistical Learning by Hastie et al.
  • Cracking The Machine Learning Interview by Nitin Su

Let me know if you have any more questions or anything I can elaborate on. Don’t worry, you got this!

#3

Hi! Thanks so much for replying with such a thought out and elaborate answer. I just had a few questions. So firstly, is it possible that I could be asked a Leetcode style algorithmic problem that’s really just a math/prob problem in disguise? Also, is there a Leetcode equivalent for these types of problems?

#4

No, I think the leetcode questions are likely to be justt leetcode questions (I’d say on the medium difficulty, sometimes hard). However, you may be asked follow up questions that involve a bit of stats or math. It is unlikely to be asked a probability question that’s a leetcode problem. But as there is still a probability (pun intended) that the coding question is related to probability, you can check out Cracking the Coding Interview by McDowell for a few probability coding questions and their solutions. For example:

  1. Write a method to generate a random number between 1 and 7, given a method that generates a random number between 1 and 5 (i.e., implement rand7() using rand5()).
  2. Write a method to shuffle a deck of cards. It must be a perfect shuffle - in other words,
    each 52! permutations of the deck has to be equally likely. Assume that you are given
    a random number generator which is perfect.
  3. You have a basketball hoop and someone says that you can play 1 of 2 games.
    Game #1: You get one shot to make the hoop.
    Game #2: You get three shots and you have to make 2 of 3 shots.
    If p is the probability of making a particular shot, for which values of p should you pick
    one game or the other?

Btw, here’s a good list of practice lc problems for ML interviews https://mlengineer.io/common-leetcode-questions-by-categories-532b301130b and if you haven’t already, you can also check out Leetcode LinkedIn (Leetcode Linkedin tag 的题目列表 - Interview - 1024 BBS).

1 Like
#5

The resources and advice you’ve shared have been immensely helpful. Thank you so much!

1 Like