You're making two assumptions - both wrong:
1) That this is an unscalable solution - A monolith app server backed by Postgres can take you very very far. You can vertically scale by throwing more hardware at it, and you can horizontally scale, by just duplicating your monolith server behind a load-balancer.
2) That you actually know where your bottlenecks will be when you actually hit your target scale. When (if) you go from 1000 users to 10,000,000 users, you WILL be re-designing and re-architecting your solution regardless what you started with because at that point, you're going to have a different team, different use-cases, and therefore a different business.