Are system design interviews easier to clear?
Your toolbox in system design is only so large. You only have a few colors and shapes of Lego bricks... additionally if you’ve done it IRL there are nuances and issues you can speak to that will greatly increase interview performance.
Compare this to an esoteric LC problem like trapping rain water 3d.... or really any LC problem. If you don’t see the intuition or the “trick” to solve the problem then you don’t clear the interview, and there is no discussion you can add to makeup for that, talk of brute force can only get you so far.
My point is, the space of system design interviews is much smaller, and the service or pipeline you will be building is probably similar to something you’ve actually built before.
Meanwhile the space for algorithmic questions is huge, everything from math and linear algebra to backtracking and bitwise ops. If you haven’t seen bitwise xor trick for example, you fail!
Don’t get me wrong, for system design, you still need to know B trees, LSMT, consistent hashing, probabilistic data structures, MapReduce, HDFS, and so on... but no ones going to ask you to implement them from scratch or really go in-depth into these. Additionally, you can always make an argument for alternative solutions.... some people still use sharded MySQL as a poor man’s Cassandra, and they have their reasons for doing that. The “wrong answer” is more tolerable.
So, now that I’ve made my case, I claim that system design interviews and more senior interviews are generally easier to clear than algorithmic heavy interviews.
comments
For instance, how many interviewers have heard of CAP theorem? Difference between quorum protocols. What context switching is in OS. And so on.
But any schlub interviewer can look at a LC hard solution and expect interviewee to get it in 30 minutes bc there’s often only 1 right answer.
System Designs are for finding the correct level to hire.
Unless its Principal level role for which there might be lot of weight on Design
This comment was deleted by original commenter.
But, I used to work at a consulting llace and the leads and architects there still knew a lot of this stuff. You’d be surprised, it was a small place but they still had Kafka/Spark/Kubernetes experience because they had big clients.
Also, doesn’t Roblox have good infra?
What I’m trying to say is: more and more places are handling big data and infra, or so it seems to me, therefore not being from FAANG isn’t that big of a deal.
But if this was 2005, and you were asked system design by Google, of course you would bomb back then. But practices and knowledge has democratized a lot since.
You overestimated what people actually do at FAANG. I would argue it is the opposite. You are more likely to be exposed with system design at startups than faang. Faang hires so many engineers and majority dont even have chance to build something big due to the amount of red tape.
There is a lot of good resources for system design but because it is so open ended it is hard to find a place with good repo. And that is not a bad thing. System design is meant to be discussion.
Also if you can't tell if someone is BSing you then that means you are not qualified to interview for that module. It doesnt mean the module is bad. While with leetcode some new grad has false impression that they know how to interview and fail someone when the candidate deviate from textbook answer.
System design interview is also more open ended and can be answered more reasonably by good candidates since it will be an open ended discussion. Leetcode is not really open ended.
Interviewer can easily throw curveball that the candidate cant easily game, provided the interviewer itself is good at it.