The nice thing about vm with nginx proxy manager or just nginx running on the same host as the rest(or majority) of vms is that internal traffic doesn’t traverse other devices. This only applies if your backend services are not configured with TLS so you’re effectively terminating at proxy and run unencrypted traffic to backend. That being said chances of some packet sniffer running on your internal network between proxy and destination VM is low.

I’m in similar situation as you. I run overpowered router that barely sees any CPU usage.

I tried Nginx opnsense plug-in but looks like GUI doesn’t support proxy by header (locations are path based). I don’t want to ssh and mess with raw config files. I’m running HA proxy on opnsense router. I saw in community forums most people use that. After going through tutorial for one service it’s pretty easy to grasp configuration concept and replicate for other services. I think only one confusing option is that backends pools and rules can have backends configured and you can have only one in use when assigning rules to public service. Test syntax button ensures you don’t make mistakes. HA proxy has powerful options for backend more than you probably need. I moved router management port to higher number and setup proxy to run on 443. Then wildcard DNS entry points to router and that allows to keep adding services as needed.