I managed a remote team for 18 years. One thing you might think about is that a lot of communication just doesn't happen naturally between remote folks. For most of that 18 years everyone was remote, we had some office space but even then it was a minority of local people and only a couple of engineers were local.
Communication is necessary between devs to make chunks of work in flight come together nicely. When you are all in the same place, you go to lunch, these topics come up, they tend to get sorted. Yes, one could argue that every chunk should have an architecture document down to the last detail, we didn't work that way, it was too slow.
When you are all distributed someone has to make sure that all the work in flight is going to come together nicely. That job fell on me as the manager and the main driver. It was a shit ton of extra communication. I had to know where everyone was and whether they were on track. Sometimes my people would self communicate but a lot of the time it was on me to know and get them to talk to each other.
By and large, it worked and worked quite well. One thing that was critical to it all was code review. We built our own system, it was a web interface to a database and repo per chunk. You could walk the diffs, comment on them, all in the web or you could clone the repo, toss in some fixup commits, push it back. The review system remembered you and had a way to show you only the diffs since the last time you reviewed this chunk. It was fairly easy to use and useful, people liked it. And it helped immensely in getting the chunks to land in the tree in good shape.
I can't say if other managers have had to pick up the communication slack for remote people, maybe that was unique to me because I was management and sort of the "dumb" architect (my people were smarter than me but they had to push stuff through my brain, much to their annoyance from time to time. The upside was that the system was pretty consistent, it was done by a bunch of people but it felt, mostly, like it was done by one guy).
If it is common for the communication load to be higher on the manager/architects when dealing with remote people you might want to think about that. It doesn't really scale, you are taking what should be a web of communication, lots of sideways stuff, and making it more centralized. My people did complain that I was the bottleneck, I don't think I could have made that model work for a much larger team. I'd have subdivide it like Linus does for the kernel.