I might even start a little closer. Take off from Texas, orbit earth a few times and land in the Mojave Desert. Have robots build his Mars city there first to prove it out. Let humans live in it for a few years. Optimize and revise.

Any kind of test environment closer to home is helpful.

Notably the Starship system takes at least 12x the launches to put a payload into deep space than it takes to put a payload into LEO.

With those economics it might sense to build large structures in LEO, such as a flashy space hotel or a simulated lunar or martian environment patterned after a very small O'Neill colony -- recent research seems to show people can tolerate a higher rotation rate than O'Neill thought so a small Bernal sphere made entirely from terrestrial materials looks possible.