1) Try to learn as quickly as possible which parts of the code are critical infrastructure and which parts of the code don't warrant perfection. I know this is probably difficult in an environment where every single character in a PR is heavily scrutinized, but time is a finite resource, and you need to be very careful not to overengineer things are not critical or will be replaced eventually. This was one of the hardest things for me to learn. Perfection is the enemy of the good.
2) Learn to love working on legacy codebases. This will happen everywhere you work. You will rarely (if ever) have the opportunity to start from scratch. If someone asks you to use an existing piece of code, and you decide to re-write it because you don't want to take the time to understand what's going on, you will A) appear to be anti-collaborative and B) reinvent the wheel, thereby wasting time. Obviously, there are scenarios where re-writing something from the ground up is necessary, but make sure all stakeholders involved agree and that you can estimate it correctly.
3) Always try to coax out as many details from designers and product managers/owners wherever possible. Perfect requirements will never exist but you want to be equipped with as many use-cases as you can before you start hacking on something. This may sound a bit waterfall-ish, but PMs and designers are not omniscient creatures. There is almost always some sort of gap that you may be able to help identify.
4) Don't be a dick, because there will almost always be someone who knows more than you. This applies to comments in PRs! Even if the entire company treats you like a god, that does not give you the freedom to treat others like worshippers.
5) If you are asked to estimate how long X will take, DO NOT give an answer until you have enough details so that you can make a reasonable estimate. If they force you to give one, pad it 2x or more depending on how much you don't know about what needs to happen for the task to be marked 'done'. Even better, ask them for a separate fact-finding task (spike, POC, or whatever you want to call it) so that this can be more easily tracked.
I could go on forever but these are some of the more non-obvious things I don't hear talked about enough.
2) Learn to love working on legacy codebases. This will happen everywhere you work. You will rarely (if ever) have the opportunity to start from scratch. If someone asks you to use an existing piece of code, and you decide to re-write it because you don't want to take the time to understand what's going on, you will A) appear to be anti-collaborative and B) reinvent the wheel, thereby wasting time. Obviously, there are scenarios where re-writing something from the ground up is necessary, but make sure all stakeholders involved agree and that you can estimate it correctly.
3) Always try to coax out as many details from designers and product managers/owners wherever possible. Perfect requirements will never exist but you want to be equipped with as many use-cases as you can before you start hacking on something. This may sound a bit waterfall-ish, but PMs and designers are not omniscient creatures. There is almost always some sort of gap that you may be able to help identify.
4) Don't be a dick, because there will almost always be someone who knows more than you. This applies to comments in PRs! Even if the entire company treats you like a god, that does not give you the freedom to treat others like worshippers.
5) If you are asked to estimate how long X will take, DO NOT give an answer until you have enough details so that you can make a reasonable estimate. If they force you to give one, pad it 2x or more depending on how much you don't know about what needs to happen for the task to be marked 'done'. Even better, ask them for a separate fact-finding task (spike, POC, or whatever you want to call it) so that this can be more easily tracked.
I could go on forever but these are some of the more non-obvious things I don't hear talked about enough.