Given that you've already noticed the utility in theoretical comp. sci., you might want to look into the quantum computing literature. Typically, you find these come up in terms of special cases of Cayley graphs rather than general Cayley graphs. (I don't know if you are interested in the particular properties of a Cayley graph, however.)
One simple example is an Ising Hamiltonian for some qubits in quantum theory. Take, for instance, the simple Hamiltonian
$\displaystyle H = -\sum_i \sigma_x^{(i)}$.
Now, consider how this Hamiltonian maps particular qubits. Specifically, if $|b_i\rangle$ is a qubit-string, then
$ -\sigma_x^{(j)} |b_i\rangle = |b_i \oplus 1_j\rangle$
where $1_j$ is the bit string with a 1 in the $j^{th}$ position and 0s elsewhere. Now, we see that
$ \langle b_i | H | b_j \rangle = -\delta_{ij}$.
This Hamiltonian can then be represented as the adjacency matrix of an N-cube acting on the set of vertices identified with bit-strings of length N where $b_i$ and $b_j$ share an edge if and only if $|b_i \oplus b_j| =1$ where $|\cdot|$ represents the Hamming-distance. We could now go and modify this Hamiltonian or the basis it acts on (ie. change it from qubits to qutrits, dits, or whatever else you choose) to cover a broader range of Cayley graphs.