Hey @Pream Pinbut
Typesense specifically is a in-RAM search engine, so all your documents - and configurations - are loaded to RAM on initialization. That increase in RAM that you see is Typesense initializating, until it hits a stabilized point. Even then, you'll see it increase and then going back to normal over time as searches comes in.
The amount on your graph matches what we see on freshly new clusters without any data (since 4 documents it's pretty much nothing)