You are correct that the boundary of the set $A=[0,1)$, considered as a subset of $\mathbb{R}$, is the set $\{0,1\}$, by exactly the reasoning you described.
However, treating $A$ as a topological space in its own right (i.e., given the subspace topology, a.k.a. the relative topology), the boundary of $A$ is empty. The same is true of any topological space $X$: the boundary of $X$, as a subset of itself, is empty.
Why? Well, the closure of $X$ in $X$ is the smallest closed subset of $X$ that contains $X$. The only subset $X$ whatsoever that contains $X$ is $X$ itself; and $X$ is closed as a subset of $X$ (since it is the complement of the open set $\varnothing$), so the closure of $X$ as a subset of $X$ is just $X$ again. Similarly, because $X$ is already open as a subset of itself, the interior of $X$ is also $X$. Thus $$\partial X=\overline{X}\setminus\mathrm{int}(X)=X\setminus X=\varnothing.$$
In short, the issue is that "boundary" is a term that depends on two pieces of information: the set we are taking the boundary of, and the "ambient" space that we are working in. The same is true of "closure" and "interior"; there is no such thing as the closure or the interior of a set. It is always relative to some "ambient" set where everything takes place (which may be left implicit, if it is assumed that it will be clear from context).