First, you don't place any restrictions on $k$. This will turn out to matter.
Second, "$f(x) = k x^{\frac{1}{2}}$" is not a function. A function presented this way is a domain and a rule/formula/expression for converting elements of the domain into elements of the image. When the domain is missing, it is normally understood to be the largest subset of whatever set makes sense given the current subject of discourse. You seem to indicate that the real numbers is that set. In this case, the domain of $f(x)$ is $x \geq 0$. (Because there are no challenges to taking square roots of zero and positive numbers.) If $k=0$, we could conceivably argue that the domain is all of the reals, but that will fail to be interesting in the next step, so we set that aside.
Now that we have a fully specified function, we can try to find its inverse. If $k=0$, we're done because $f$ is not injective (equivalently, does not pass "the horizontal line test") so does not have an inverse. (If, for some reason, the domain of $f$ had been a single point, we would be able to continue -- the domain of the inverse will be the output of $f$ at that one point.) Otherwise, if $k>0$, the range of $f$ is $[0,\infty)$ and if $k < 0$, the range of $f$ is $(-\infty,0]$. Using the fact about functions and their inverses you mention, the domain of $f^{-1}$ is either $[0,\infty)$ or $(-\infty,0]$ as $k >0$ or $k < 0$, respectively.
We could go a bit further to check the above. We compute \begin{align*}
f(x) &= y = k x^{\frac{1}{2}} \text{, so} \\
f^{-1}(y) &= x = \left( \frac{y}{k} \right)^2 \text{.}
\end{align*}
If $k>0$, we have $y \geq 0$, and we are looking at a point on the same half of the square function that becomes the upper half of the graph of the square root function under inversion. If $k < 0$, we have $y \leq 0$, and we are looking at the same half of the graph of the square function that becomes the lower half of the square root function under inversion (as we must since $k<0$ means we are only considering that half).