The take home exercise is said to be a code review. It is not. It is a strict coding challenge with a predetermined grading rubric - they will look for and judge precise code solutions, which you translate into the form of comments on a fake PR. You have to download, run, debug, fix and test ~700 lines of intentionally awful code (React and SCSS). Then you transfer all your findings into HackRank and do your pretend code review. I commented on 34 lines and I was holding back to avoid coming across as an asshole. Recruiter said I got a very bad score😒 The Slack blog says it takes "at most two hours to complete". Recruiter said 1-3 hours. It does not. With no context you go into HackRank and get 4 hours to: - read a bunch of instructions - get familiar with the mock UI design - download a React project (cautiously ignore the pop-up warning you that reviewers will know you've left the page and to not do that. wtf?) - set up this project and orient yourself with the files - run it - find and debug the bugs - fix and test - then write it all up as comments line by line in HackRank as the 4 hour timer counts down - then proofread and debate whether you should keep going or show mercy... Hope you didn't start this thinking it takes "at most two hours to complete" and now you've got to push out the rest of your calendar or try to rush and cram in a "solution". Whatever. The bizarre thing is *none of this reflects reality at all* and Slack 100% knows it! This pretend code review they make you do breaks from every bit of *Slack's own guidance on what makes a good review* lol (see https://slack.engineering/how-about-code-reviews/) Are new Slack hires assigned a dumpster fire PR on day 1 and given 4 hours to complete it or be fired? Ok, so why is this the interview? It's the exact opposite of reality: - not just reviewing code - fixed 4 hour time limit - not an actual PR, i.e. not in actual repo with none of the helpful tools we use every day doing real code reviews - zero context to work from - ridiculously bad code, and lots of it - waste everyones time by leaving 30-40 comments attempting to fix all of this guy's broke code - there's no real human teammate to talk to, e.g. [ironically using Slack] "Hey dude, took a look at the PR and it needs work. Let's huddle and we'll go over it together." In short, you have 4 hours to do the most ass-backwards code review/code challenge you've ever done. 🤦♂️ #interview #slack #mystery
Can you use chatGPT to solve it?
you should 100% .
ChatGPT can give wrong answers now. It's not much helpful.
Still better than leetcode
It's not because it is a leetcode! It's a leetcode inside a feature inside a code review. You have to address all three of these layers. So you're not just solving a challenge and trying to do it elegantly, you're doing that AND reworking things to be more "correct" according to The React Way AND address a dozen other little broken/janky things AND build a portion of the feature that is completely missing (I just politely mentioned it was missing cause no one on planet earth does a code review and goes and builds out totally missing requirements) AND fix broken Jest tests AND, and this is one part I ignored, apparently clean up all the crap SCSS too.
They gave you feedback — you lack attention to detail. Given this is a frontend loop, you’re going to have to have an understanding of everything including SCSS (or some variant)
Can you refer me
😂😂😂😂
Brain is looking to move out of Coinbase
Are you kidding me? This sounds great. Should take 10 minutes tops to write a single comment saying the PR author is clearly not ready to contribute to the codebase and should look into alternative careers.
🤣🤣🤣
“I have some concerns about this approach, lets schedule a sync to talk through it”
I did the same at salesforce.. I appreciated it way more than a dumb leetcode. I interviewd @ Tesla autopilot and cloud kitchens team few years ago.. the take homes were the best in class. You build an actual product.. from that they get to test your coding skills, practices, code smells, your logic, did you write test. Second round would be to go over the code and do a group review.. understand how you would have scaled it, what you would have done differently if time permitted etc…
That sounds way more insightful, realistic, productive, and fair. This Slack thing was so incredibly unrealistic imo. A naive mid-level engineer might attempt to do a code review on a garbage dump like that, but this was for a Staff role. I have done 1,000s of code reviews over the past 10+ years and the correct answer here is to NOT waste cycles attempting to unravel all this with 100's of comments in the PR and instead use it as a teaching moment face-to-face with a junior to help them learn how to write better code. So it's just really dumb they frame this as a "just a code review". Give candidates the same requirements and ask them to build it and leave lots of comments for the reviewer. This unrealistic step of reworking the 700 lines of crap into something good via PR comments is just retarded.
And how exactly do you go about that? Is there some way, obvious to staff level, where you can go over X amount of things briefly in a face to face / huddle, vs 10X the amount of comments? I’m reasoning that the huddle does not need to happen, but instead writing out all those comments would itself count as a one time mentoring session with this particular engineer. Unless of course I’m missing something.
For a company called Slack
What’s wrong with slack?
Oh and you are in what?? Wells Fargo? Give me a break!
Dude, you should have just used "LGTM" and they would have hired you.
🤣
🚢🚢🚢🚢
Git gud noob
Why would I want to work somewhere I have to review, fix, and rewrite someone’s buggy code?
Because this happens at every company
Somehow hasn't happened to me in about five years, at my current employer, though, yeah, it happened a few times in other jobs.
Classic, I have not seen a single decently large company have any type of reasonable assessment strategy for hiring. I think literally pulling randomly from the resume pile will yield similar results compared to what the unicorns and FAANGs are doing.
Might lead to better results imo
apply to ramp it’s different