When we take a derivative of the function f(x) with respect to x we find out how much an infinitely small change in x will change f(x).
When we take the gradient of a multivariable function say f(x,y,z) we define it as
$$ \nabla.f(x,y,z) = \frac{\partial f(x,y,z)}{\partial x} \hat{i} + \frac{\partial f(x,y,z)}{\partial y} \hat{j} + \frac{\partial f(x,y,z)}{\partial z} \hat{k} $$
and if the position vector is given by $ \vec{r}(x,y,z) = x\hat{i} + y\hat{j} + z\hat{k} $
Then is it safe to correlate that the gradient is the directional derivative of the function with respect to the position vector $\vec{r}$
If I am wrong can you explain in detail why ?
No, because the directional derivative of a function does not change it's "type", e.g. the directional derivative of a scalar-valued function is a scalar-valued function, not a vector-valued function.
But choose any vector $\vec v$ and calculate $v\cdot\nabla f$. You will find that this is the directional derivative (and perhaps you should review the definition of directional derivative). So the quantity you want is $\vec r\cdot\nabla f$; but note that $\vec r$ does not get differentiated in any way.