I am always surprised by devs that have a strong sense for O notation and how it affects code complexity, but have no clue about the bloat in tools or libraries

I am always surprised by devs that have a strong sense for O notation and how it affects code complexity, but have no clue about the bloat in tools or libraries they throw on the pile. So version 1.0 comes around and they immediately need to refactor to get down from a 15 second page load (or from a huge file/data stream). Fine tuning code is one thing, but installing a bloated package for one tiny feature should be well-considered at the time of implementation because every addition adds to the tech debt bill.