What’s better for improving at real system design? Too early?

For real-world architecture/system design, earlier in your career, is it better to learn more passively first by taking your time learning from others/attending meetings, or by actually doing it and actively participating from the start? Is good feedback/mentorship what truly matters or is it feasible to get good via self-teaching/research/testing? I have the option of being on a startup-y team with more responsibility vs. a mature env where I can access more resources/existing processes but likely contribute less. I’ve gotten my feet wet and want to level up my design skills, but I’m just mid-level with 2 YOE. Is it too early/ambitious to try making decent architecture decisions at 2 YOE (what about 1 or 3)? Is a scrappier env typically detrimental to good sys design? And this can apply to different domains (frontend, backend, mobile, etc.). There are 2 pairs of choices in the poll, so there can be multiple selections. TC: 160k #systemdesign