It all comes down to if I can solve a problem like “Given an NxM matrix of integer strings, print true if any row, column, or diagnonal contains only palindromic numbers that can be written as the sum of consecutive squares.“ Oh, and on some random guy’s laptop, with a panel of 4 people watching me type as they sit in silence, and with 20 minutes remaining.
When I look back at all of my career successes (leading engineering teams through 10 successful commercial products, 4 startups, 3 acquisitions, millions of users, etc.), coding was hardly important.
I guess it’s time to retire.
Second, once you get past a minimum level of intelligence a lot of success in software development is based on motivation. It's hard to be motivated by a made-up problem where all your hard work is going to be thrown away and not used by anyone for anything, and my first step in solving it is to involuntarily roll my eyes and groan. That starts the project off on a bad foot to begin with.
You not liking them doesn't make them stupid.
I once posted a puzzle on a different website that boiled down to "why did I choose this particular non-round number" with some context, and eventually someone got it, but some people got very angry at me.
The trouble is that a lot of puzzles reduce to "can you follow a particular convoluted path to a solution" and if you don't, you are being judged unworthy for being dissimilar in your thinking and not necessarily for your sheer brainpower.
That's why I prefer puzzles that nobody has solved, but those have two obvious problems - either (a) they are too hard to solve, or (b) once solved, they appear to be too easy. Either way, you can't use them as tests.
It's natural to like tests that you are good at and other people are not, and to hate tests that other people are good at and you are not. So, people are always going to take offense at tests.
Ok, but this is not really a puzzle. It looks like a rather trivial exercise just to see if you can write nested loops. I guess an average software engineer shouldn't have problems with this. Or I am missing something.
I don’t know why I can’t do it. I freeze up with analysis paralysis, or stage fright, or something. Maybe it’s because I have no clue what they’re looking for, and I’m so used to simplifying problems based on business cases and stakeholder feedback. Maybe it’s because I’m always thinking about production-grade code (I do try to avoid nested loops!). Maybe because I assume that the problem requires a specialized data structure, and so try to invent it on the spot. Or, it could be that I am old-school and like to solve problems on paper first before coding.
You may not believe me, but I code everyday. At every job I’ve held, I’ve been complimented as one of the best programmers. So, maybe I’ve only ever worked with C-grade developers? But, I’ve worked with 100s of people. How can that be?
But, let’s say that I really do suck at coding. Are all my other attributes, skills, and experiences valueless? Seems like such a waste.
Years ago, relatively early in my career, I went into management. But, I only did it for 2 years before going technical again. It just didn’t make sense at that time. Senior developers then made more than managers, and I enjoyed building things. Now, I really regret that decision.
Anyway ... I clearly can’t get a job. Anyone have any career ideas for a washed up software engineer?
This is an interesting list that has been posted here before:
- https://github.com/poteto/hiring-without-whiteboards
But if you want to work somewhere firmly in the whiteboard club - do regular practice on hackerrank and leetcode. Many interviewers will draw their questions directly from those sites (and some others, I'm sure).
(Personally, I honestly feel those exercises have improved my coding, so there's that too.)
I thought it was kind of silly but have you tried practicing on hackerrank or leetcode for the coding tests? Hackerrank feels pretty brittle but a lot of companies use it for timed screening so good to get a feel for the platform's ability to lose all your work in a non pressure situation, leetcode has very useful discussions about user supplied solutions.
You're not alone. I recently went thru this myself and wrote up my experiences here: http://fuzzyblog.io/blog/jobhound/2018/04/24/ten-things-i-le...
I hope that helps.
Also, there are strategies for managing the time problem. 20 minutes is a bit fast for understand => ask questions => implement => test on this problem. But you can organize the code so that if you run out of time it's still clear you knew how to solve the problem and can code your way out of a wet paper bag. E.g., I would pass an answer that got everything correct but didn't get around to implementing isPalindrome() or isSoCS() before the 20 minutes was up, especially if the candidate could give a ~1 minute description of how to implement those functions. (I would also give more than 20 minutes.)
However, I haven't experienced repeated failure of a consistent type, so I'm not particularly motivated to worry about this sort of thing. I do know that different companies have radically different processes, so if someone tells you "you're not a good fit" it may be completely true and not a negative judgement at all (even if that's how they mean it).
It so happens that I got my undergraduate degree about ten years later than is typical, and thus I can pass for someone that much younger than me, at present. So in an interview, it is tempting for me to give away my age by reminiscing about computing in the 80s, and that could have caused me to be rejected a couple times but I will never know for sure...
But a surprising number of applicants fail to pass a simple coding test, not even getting a naive solution.
One thing that really helps with perspective, whichever side of an interview you are usually on, is to have seen how radically differently different companies do things.
That doesn't necessarily mean that you should change your practices, but diversity in behavior exists because there are niches that can be exploited and trade-offs to doing so.
That part, to me, is worse then the question. Are you exaggerating, or did the panel really just sit there and watch you? If so, can you say which company so people know to avoid?
The first kind of company will not underestimate certain workloads and pays experts very high salaries. The second company really doesn't need to succeed technology wise, so they use the cheapest way possible to create their fluff.
And one tip that always applies: Never tell your manager about your ambitions. Never! A middle manager is certainly someone who is more about the fluff+constant pay type of person, a rent seeker. He doesn't like people showing true effort and actually gaining leverage for raises and maybe even take over his position. Especially the last point is very important to him. For him his position is the most valuable thing in the world so it is natural from his point of view that everybody below him wants to steal it from him. And it's also not totally wrong since he probably got his position just in this way. So even if you work for Amazon who really want to compete for AI leadership don't tell your manager. Maybe talk with HR or do it in private, and then apply for another team when you got the skills.
For me I'm open and direct with managers. I've had some good ones and bad ones. Id say 70/30 ratio in favour of good. And when a manger is bad the adage plays true, 'People do not leave bad companies, they leave bad bosses'.
a) He doesn't want to move up. The higher you get the more visible lack of skill and ambition becomes as well.
b) “It is not enough [for him] merely to win; others must lose.” ― Gore Vidal
Sorry, that's all I got about this question.
I definitely use a lot of statistics daily. I do not need to compute integrals by hand or construct formal proofs, but a lot of intuition is necessary.
Statistics is very tricky by the way. It is so easy to have incorrect assumptions and totally misapply the methods. Also another problem with ML in general is that if you make an error you very often do not notice it. The methods will usually work a bit worse but they will still kinda work.
Just yesterday I discovered a tricky feedback loop that corrupted my data for several weeks.
I think this why HR hire computer engineers instead of statisticians when it comes to ML.
The vast majority of applications for AI/ML can be done with libraries, just like the vast majority of programmers doesn't need to understand how to make a compiler or even how memory works. It just helps.
At some point you need to explain why something works.
A random 40 y/o programmer might be rusty on it, but it's not going to take very long to pick it all up again.
And if you've done any game programming even just for fun, you probably know quite a bit of that already.
Bad way to go about it anyways. The best way is to change companies if possible (granted, one will have to put in their own time to learn or do it quietly at their current job). If the company is good about intra-company movement, it might be possible to do this within the company, but often the companies that can do this come off as too large to have the flexibility one needs to really dive in as opposed to getting lost in the fray (you'll feel like you're a junior all over again). Unfortunately, mobility is the antithesis of stability.
I would claim that the more general and reusable the tool ("AI platform") is, the more experience is needed to build it.
I'd expect the output to be logarithmic.
However, he will not do exactly the same thing. For instance a 10+ years guy might realize that instead of investigating a topic for 3 workdays, he can ask the colleague from another team he worked with 2 years ago, and get the answer in 20 mins. His processes will change a lot, so his output can increase exponentially even. Of course it's limited by his intelligence, i.e. his ability to learn from his experience. And it's likely to happen in steps. So a quite smooth line for a few months, then an aha-moment where he bumps his output drastically, then another smooth line.
A developer can only type so much each day, after all.
The manager said that he could "gather a bunch graduates" to build an AI/ML platform. There was no need to experiment and bet on someone with that much salary and experience.
So, I somehow feel ageism in tech is the result of everyone thinking tech is easy to build. A 15+years JAVA guy has similar output to say 5-6 years or for some extreme managers even 2-3 years.