There are a couple of ways to do this. The first one relies on the ESP booting up and pulling the pin low before the RC voltage rises enough to turn on the NFET. This may affect the bootup behavior you talked about though.

simulate this circuit – Schematic created using CircuitLab
The other way would be with a PFET and BJT. The base and emitter are both pulled up, so the transistor is off at power up. Then when the ESP is ready, it drives the control pin low, which pulls the emitter low and turns on the NPN transistor, which then provides a negative bias Vgs to the PFET to turn on the load.
