Consider the following identities
\begin{align*} \sum_{k=0}^n\binom nk x^k(1-x)^{n-k}&=1\tag1\\ \sum_{k=0}^n\frac kn\binom nkx^k(1-x)^{n-k}&=x\tag2\\ \sum_{k=0}^n\frac kn\left(1-\frac nk\right)x^k(1-x)^{n-k}&=x(1-x)\tag3 \end{align*}
There are quite straightforward to prove by first using the Binomial Theorem for $(1)$ and then deducing $(2)$ and $(3)$. However, we can obtain the following chain of equalities
$$\small\sum_{k=0}^n\left(x-\frac kn\right)^2\binom nkx^k(1-x)^{n-k}=\frac1n[x(1-x)]=\sum_{k=0}^n\left(x(1-x)-\frac kn\left(1-\frac kn\right)\right)\binom nkx^k(1-x)^{n-k}$$
In particular, therefore we can deduce that
$$\small\sum_{k=0}^n\left(x-\frac kn\right)^2\binom nkx^k(1-x)^{n-k}=\sum_{k=0}^n\left(x(1-x)-\frac kn\left(1-\frac kn\right)\right)\binom nkx^k(1-x)^{n-k}\tag{$\star$}$$
This is where the fun begins! Is it somehow possible to obtain $(\star)$, without relying on our intermediate chain of equalities? Playing around with the two sums I cannot see a direct way to attack the problem and honestly I do not know what else to do.
Is it possible to show $(\star)$ without actually evaluating both sums?
Thanks in advance!
Looks like I have overlooked something quite trivial as cross-posting the problem on AoPS attracted a simple solution by computation, which I will post here for reference. All credit due to the user ysharifi.