A metaverse engine server (probably a whole farm of services) is somewhat like a MMO game server, except that it may have to deal with other similar servers. Possibly untrusted servers run by others, if you allow portals or teleporting or adjacent regions run by different grids. Hard problem. Improbable tried to build such a thing, as Spatial OS, but the end result was really expensive to run and all the games that used it went broke.
Both Tim Sweeney and John Carmack see many unsolved problems in this area. If this "metaverse" stuff survives the hype and the scams and turns out to be important, there's plenty of hard work ahead in this area.
Plenty of engine companies and studios will hire juniors and mid-level engineers into engine teams. And it’s been that way for a long time. And whilst the engine will preferably be made whilst developing actual games using it you don’t need everyone on the team to have been a gameplay programmer. There’s almost too much specialisation required these days and a lot of engine engineers aren’t interested in or good enough at game design to do gameplay roles. What it does need is active feedback from game teams building serious projects.
I’d agree though if you want stability and much less chance of crunch pick a company with a stable income. Engine companies are good for this but there are plenty of others.
Yet there also exists a sea of inexperienced programmers jumping straight into "engine dev". They have none of the critical production experience which drives quality engines. This flood of enthusistic yet unprepared engineers causes a bipartite population. With the majority of public articles or YouTube presentations coming from those least prepared to speak with knowledge.
If you want to become a game engine dev your first step should be joining a studio as a gameplay programmer. No one wants to use an engine written by someone who does not know how to make games.