Such a heater would depend on the dielectric loss
@ frequency of the material. Milk or cheese may or
may not have an absorption peak in the stated
frequency band, and the exact frequency center
of that peak (if any) is important.
I'd begin with chasing down references on that
(RF absorption properties of dairy products) and
then worry about getting the correct frequency
and power out of some black box or other. If there
is no strong peak then another heating method
may be a better bet.
I imagine there will be other interests such as
sample agitation, etc. - like in a microwave oven
(which works on water's absorption peak in the
2GHz range), absorption occurs preferentially at
the outer surfaces and attenuates the energy
available to the interior. 10kW might just give you
a crispy shell around cold glop (or something).
Maybe applying power offset from frequency peak
could produce a more uniform heating, but that
too wants some checking of present art.