Google SWE Interview Preparation

I interviewed at Google Pittsburgh a while back (as a result of Google FooBar), and while I signed an NDA regarding the interview questions, I can provide a brief overview of the process. Ultimately, I did not receive an offer, so take this for what it’s worth.

Preparation

Google will email you some official interview preparation materials, which you should obviously review. They outline the process very thoroughly, as well as provide an outline of possible material. If you’ve prepared for technical interviews before, much of this content is not a surprise, but it would be foolish not to review everything they’ve sent.

How does their interview process work?

Typically, there are phone interviews, then an on-site interview. I skipped the phone interview stage because of FooBar, and went directly to the on-site interviews.
If you are selected after submitting an application, or re-apply, you’ll be asked to do a phone interview first.
Since I can’t offer guidance here, I’ll refer you to Google’s Interviews page for specific details.

How much time should I allot to studying?

This answer depends on how comfortable you are with your CS fundamentals. Most people dedicate at least a month, possibly more. A recruiter told me they’re not able to schedule interviews greater than 30 days ahead, but you have the option of contacting them later to schedule. From every interaction I’ve had (a couple recruiters, and the engineers on-site), they genuinely want you to be at the top of your game when you come in. Take your time. They’re almost too cool about making sure you’re prepared for the interview process.

On-Site Interview

The on-site interview can be done over one day or two. I’m not sure what game theory says here, but I went for the one-day interview. This consisted of five interviews, about 45 minutes each. (You’ll also meet up with another engineer for lunch, which isn’t really part of the interview process.) They’ve even put up an example interview on YouTube:

Do not expect them to ask about your past projects, resume, etc. I saw a lot of complaining on glassdoor about this (mostly from people who didn’t get an offer).

They’re less interested in your specific background and accomplishments than your ability to solve the problems presented, which seems to offend a lot of people. Furthermore, everyone I met was super friendly, except for one interviewer who really didn’t seem interested in stepping away from work to interview someone. I’m told this is happens most frequently in phone interviews, though.

Generally speaking, the problems I was presented had a brute-force solution and an elegant solution or two. If you reach a working solution, they’ll likely ask a few cursory questions about Big-O notation or what data structure you’re using, then ask you to iterate on your code to meet additional requirements, consume fewer resources, or otherwise refine your solution. While they might appear to be tricky questions, they’re really not out to get you. The problems are very much in line with the TopCoder Division I problems, and I’m told that being comfortable with solving those types problems correlates with success at Google.

I was able to solve two of the problems relatively easily, had difficulty with the third, and did not reach a working solution for two other problems. You are not necessarily penalized for not reaching a solution, but it obviously helps. I’m told they’re more interested in your thought process and approach than getting a working solution.

Review Comp-sci fundamentals

You should be comfortable discussing the various types of sorting algorithms, BFS/DFS, tree and graph manipulation, etc. You will be expected to talk intelligently about Big-O notation and discuss the running time and space constraints of the algorithms you design. You should be able to digest the problem and find the most appropriate data structure (array, stack, linked list, graph, etc). I did not have any problems that involved crazy complex algorithms or cutting-edge research.

Data Structures

  • Stacks & Queues
  • Binary Trees
  • Trie-Trees
  • Graphs

Algorithms

  • Sorting
  • Tree insertion, manipulation, and search
  • Stack/Queue problems

Practice, practice, practice

Commit to doing at least one practice problem each day. You will be expected to do one interview in a compiled language (C++, Java, or Go), but are permitted to do the rest in a common language of your choosing (e.g., Python). I’d venture to guess that nearly all Google engineers are polyglots, and as long as you’re not using Lisp or Prolog or something, you should be fine. Talk with your recruiter, or attend the prep session for answers to specific questions like these.

What Libraries Are Permitted?

Neither myself nor my interviewers were aware of a specific list of libraries that are allowed, but I was permitted to use common Python and C++ libraries (bisect, std::vector, etc.), as long as they didn’t solve the problem outright (e.g., Python’s sorted() function). You are not expected to implement everything from scratch either – they want to see modern, idiomatic programming.

Example Google Interview Questions

The internet has some specific interview questions that others have asked, but obviously Google’s engineers aren’t dumb, and Google itself is uniquely aware of what content people are searching for. They routinely change up the questions, and I’m told their validated question pool is sufficiently large that you can’t study the test. That being said, the questions I’ve seen online accurately reflect the difficulty level of the problems I had, but my problems were 100% unrelated. Be able to apply the basics.

Summary

As someone without a CS degree, the questions that I had weren’t entirely outside my grasp. I could almost see the right solution, but wasn’t quite able to implement some of them. More preparation would have definitely helped.

Many of the questions were related to problems I’d solved while practicing. The best I can do in describing them is this: they’re standard comp-sci problems, with a twist. They’re close enough to standard problems that they’ll expect you to use the appropriate algorithms and/or data structures, but modified slightly so that you’ll have to actually understand what’s going on. Rote memorization of quicksort, mergesort, etc. won’t do.

Google SWE Interview Preparation

Leave a Reply