By 0.35um you should not be assuming a long channel
simple MOS model anymore, and would be caring about
the various mobilities and their fitting companions like
velocity saturation, etc.
Maybe you want to look for PDKs at some of the university
sites (like cmosedu.com; there are others) that are good to
go.
kp/kn would be back-figured from the Id=k'(V-VT)^2 equation
but that's way too simpleminded for a good fit in submicron
technologies (bent particularly by Vds position and short
channel effects).
Well, k' = u0 * Cox in the simple models, but those may
be no easier to come by (accurately / specifically; first
principles can get you "close" on Cox, the surface mobility
is all about the gate oxide "finish" and you would have to
pull data and fit, which seems to not be your plan.