In quantum mechanics, the canonical commutation relations $[q,p]=i\mathbb{1}$, $[q,\mathbb{1}]=[p.\mathbb{1}]=0$ (I am taking $\hbar=1$) between position ( $q$ ), momentum ( $p$ ) and identity ($\mathbb{1}$) operators form an algebra, that is the Lie algebra of the Heisenberg group.
There are problems of domains because the operators are unbounded, so usually the commutation relations are rigorously expressed in the exponentiated form, the so-called Weyl relations (the blue boxed relation in the link).
Anyways, the Stone-von Neumann Theorem asserts, roughly speaking, that all the unitary irreducible representations of the Weyl relations (i.e. of the Heisenberg group) are all unitarily equivalent, and equivalent to the representation where one object of the algebra is the operator in $L^2$ that acts as multiplication by the variable, another is the derivation with respect to the same variable multiplied by $-i$ (i.e. $-i\partial_{(\cdot)}$) and the last is the identity operator.
The usual "Schrödinger position representation" is the identification of the Hilbert space of QM with $L^2$, where the position operator is the multiplication by the coordinate $x$. It follows that the momentum is the derivative $-i\partial_x$ (because they have to satisfy the canonical commutation relations), and that they form a representation of the algebra of the Heisenberg group.
The momentum representation instead is the representation of the Heisenberg group on $L^2$ where the momentum operator is the multiplication by the coordinate $k$ and the position operator is the derivation $-i\partial_k$.
As requested by the Stone-von Neumann theorem, the two representations are unitarily equivalent, and the unitary transformation that links the two is the Fourier transform (that is unitary on $L^2$).