In the simplest case, a point mass or particle or any other suitable abstraction at ($\vec r$) moving (at velocity $\vec v$) through space (with an origin), one can define an angular velocity about the origin:
$$ \vec \omega = \vec r \times \vec v $$
where the cross product's three components are defined by:
$$ \omega_i = \epsilon_{ijk}r_jv_k $$
What's that? That's the same as defining and antisymmetric rank-2 tensor:
$$ \omega_{ij} = r_iv_j - r_jv_i ,$$
which has 3 independent components that transform under rotations just like an ordinary vector.
Under reflections (aka coordinate inversion, aka parity transformations), the angular velocity does not transform like a vector:
$$ \vec r \rightarrow -\vec r$$
$$ \vec v \rightarrow -\vec v$$
(that is, vectors are odd), while:
$$ \vec\omega \rightarrow +\vec\omega $$
The angular velocity "vector" is even, just like a rank-2 tensor. It is for this reason that it is called an axial vector.
Sometimes "axial-vector" is considered synonymous with "pseudo-vector", but their is a distinction: pseudo-vectors depend on the origin.
If I translate the orgin:
$$ \vec r \rightarrow \vec r + \vec a, $$
then $\vec \omega$ changes. Real vectors, like $\vec v$ and $\vec a$, don't do that. Of course, that leaves $\vec r$ out in the lurch, because it's not really a vector either, since it has an orgin which breaks translation symmetry. Really $\vec r$ is an affine point, and in any serious physics formula you're always talking about $\vec r - \vec r'$, which is a vector.