Hacker News new | ask | show | jobs
by nineplay 1073 days ago
The problem is that the system design interview somehow became a necessary component of the FAANG hiring process.
6 comments

FAANG and similar companies typically subscribe to something like the "T shaped engineer" philosophy. They're making a conscious choice that their engineers should be comfortable in discussions about distributed systems, performance tradeoffs, etc. regardless of whether they do such things on a regular basis.
Certainly not at the FAANG I work at. We hire specialized engineers to work on device drivers and OS kernels and absolutely do not ask them questions on how to design distributed web services.

I encourage you to apply: https://www.apple.com/careers/us/

Why would you interview for a role at a FAANG company in the first place?
They want to exchange the most money possible for their labor ?
This isn’t true, and even if it were, “most money possible” isn’t a meaningful metric.
So what companies pay on average than FAANG[1] for developers.

The most money possible is far from meaningless. If I work for a company, which one will deposit the most in my bank account in a year and/or the most in my brokerage account when my stock vests?

[1] not literally FAANG, the most profitable public tech companies

Most Fortune 100 companies are competitive now, and the most money is extremely meaningless.
I assure you that the other Fortune 100 companies are not paying even in the same ball park as Facebook, Apple, Amazon, and Google.

https://finasko.com/fortune-100-companies/

What do you think the average compensation of those companies are?

I’m well aware of the comp levels at at least three of those companies because they are based in my former home town - Delta, Home Depot and Coke.

They pay their senior devs about the same amount as an intern I mentored got as a return offer (cash + stock).

They pay a lot?
In retrospect it was a horrible mistake.
They do some impressive stuff
Salary, I suppose?
My understanding is that this is not the case any more for the more junior software engineer positions in Google and in Amazon, which are expected to them learn system design before being promoted. If you are applying to a more senior position, then yes, there should be a question about system design, and yes, you will probably be doing system design in your work, so it's completely fair game.
And the second part of this is that just as all non-rich people tend to consider themselves as soon to be millionaires temporarily down on their luck, most startups (especially the VC funded ones making big promises to investors) tend to consider themselves soon to be FAANGs temporarily in the early phase of their inevitable hockey stick growth.
Is this a problem? I would argue that this style of interviewing is much more relastic to day-to-day activities than leetcode
You'd be arguing wrong imo. No one sits down solo and has to design a system to scale in isolation,and if you do then something up the chain from that moment went very wrong.

It's a pointless academia by proxy situation that encourages filling out teams with the kids of people who csn architect and tinker forever but have no capability to actually deliver software anyone wants to use. This becomes more clear when you look across the last 10 years in FAANG and list out what products have actually been delivered that are improvements to users and customers vs what's just infrastructure padding and bought in though acquisition.

In both FAANG jobs I had I was expected to design systems solo, and then review them with my team. If the system is complex enough, I would probably whiteboard it first with some teammates while I was designing it.

It is something that was asked of me in interviews, and comes up often in my day to day job. And being able to design systems, and to help review systems others are designing, is probably the single biggest impact thing I do regularly.

It is more useful in day to day than the algorithmic knowledge that was also asked of me during interviews. While there are people that do use complex algorithms in both companies, most software of Google is converting a protocol buffer into another protocol buffer, and in Amazon is the same thing but with JSON. If you are a frontend engineer, you might convert into HTML by plugging the values into a template engine.