As you can infer from christop, we have OS kernel developers among the regulars (or ex-regulars: haven't seen one of them in awhile) here. But writing one from scratch to be fully fledged and comparable with something like Linux is a different story from building on existing work as part of a team, possibly supported by a community of contributors and testers.
Having said that, of course plenty (the vast majority?) of community projects start out as one-person projects, but perhaps LoneKernelWriter's method from post #1 of this thread of attracting team/community members could be improved.