It's fun to start new projects because it's easy and there's quick payoff. That said, no one really has a place for a serial project-starter on their team and the majority of your work in your career will be like this. Learning how to explore and modify/test a large existing codebase is crucial not only to how we work on software, but also to learn how to build (or not build) new projects from scratch in the future.
That's not to say there aren't codebases that are easier or harder to learn, shops with crappy tools, or that there aren't awesome startups doing new things all the time. If you want my advice, I'd say stick around for at least a year to eighteen months and learn how to navigate and work in a big codebase with tech debt and legacy, because when you move on to start a new project you'll know how to avoid the pitfalls or how to deal with it as it arises in that project.