I might be in the minority, but I get more excited about the idea of maintaining/working on some creaky old legacy code base than I do about the idea of starting a new project from scratch.
I've gotten to spend some time where my major responsibility was to refactor and improve "research-grade" code from some scientists. Felt like tending a Zen rock garden, but code lol, I found it really relaxing and lovely.
Used to do that when I was working in science. I also kinda loved it. Just interesting to intimately experience how people thought back in the 80s. There are surprisingly many Fortran 77 libraries still in use today (they can be called from modern Fortran code).
don't worry, it takes atmost three months for that fresh code to become legacy code bogged down by decisions done in anticipation of things that never happened :)