Don't these roots sum to $0$? Then how do I evaluate this sum? The question is: $U$ denotes a set of complex roots to $a^{1020} = 1$ and $P$ denotes the set of complex roots to $b^4 = 1$. $K = U - P$.
The equation is $$ \bigg\lvert \sum_{y \epsilon K}^{} \frac{1+y}{-y^3+y^2-y+1}\bigg\rvert^2 $$
Here's a somewhat more pedestrian answer than the very elegant one given by Bran Moehring. Although the underlying ideas are essentially the same, I've used more tedious (and unfortunately, therefore, more error-prone) calculation in place of some of the ingenuity needed to come up with his answer.
Since $\ K=\left\{ e^\frac{(255m+n)2\pi i}{1020}\big\vert\,m,n\in\mathbb{Z},0\le m\le3, 1\le n\le 254\right\}\ $, then if you put $\ r_{m,n}=e^\frac{(255m+n)2\pi i}{1020}\ $, and use Don Thousand's observation you can write your sum as $$ S=\bigg\lvert\sum_{m=0}^3\sum_{n=1}^{254}\frac{\left(1+r_{m,n}\right)^2}{1-r_{m,n}^4}\bigg\rvert^2\ , $$ which can be further simplified by using the facts that $\ r_{m,n}^4=e^\frac{2\pi in}{255}\ $, $\ r_{0,n}^2=r_{2,n}^2=-r_{1,n}^2=-r_{3,n}^2=e^\frac{2\pi in}{510}\ $, $\ r_{0,n}=-r_{2,n}=e^\frac{2\pi in}{1020}\ $, and $\ r_{1,n}=-r_{3,n}=e^\frac{2\pi i(255+n)}{1020}\ $, giving \begin{align} S&=\bigg\lvert\,4\sum_{n=1}^{254}\frac{1}{1-\rho_n}\bigg\rvert^2\ , \end{align} where $\ \rho_n=r_{0,n}^4=e^\frac{2\pi in}{255}\ $. Now observe that $\ \rho_{255-n}=e^{-\frac{2\pi in}{255}}=$$\frac{1}{\rho_n}\ $, so $\ \frac{1}{1-\rho_{255-n}}=\frac{1}{1-\rho_n^{-1}}=-\frac{\rho_n}{1-\rho_n}\ $, $\ \frac{1}{1-\rho_n}+\frac{1}{1-\rho_{255-n}}=1\ $, and the sum becomes $$ S=\bigg\lvert\,4\sum_{n=1}^{177}1\,\bigg\rvert^2=508^2\ . $$