There are two downsides to this. One is that beginners glean diminishing returns from writing another FizzBuzz level program. The second downside is that the problems are not FizzBuzz adjacent.
FizzBuzz transforms a known one-to-hundred constant into a one-two-Fizz-four-etc. constant.
Leetcode problems involve engineering a program to handle arbitrary inputs with consideration of time and space complexity. That's what makes Leetcode useful for screening employment candidates. Leetcode is designed to break FizzBuzz level adjacent code.
Or to put it another way, there are not "basic Leetcode" problems because solutions are tested against "adversarial" input.
To me, it is only worth doing LeetCode problems if you are focused on handling arbitrary IO with time and space efficiency. Only worth doing if you are approaching the problems as engineering problems. Good luck.
web dev at an average to good company? no absolutely not worth it
anything at a FAANG? +/- mandatory
anything that involves complex algorithms?(efficient delivery/dispatching system at scale, complex social media, complex games, anything where efficiency/speed/low memory usage is extremely important) probably yes
otherwise some basic algorithm solving is probably enough(I really enjoyed the algorithmic exercises on freecodecamp web path for example, I think it's just enough for about 80-90% of junior programming jobs)