In QFT, there is a vector field of electromagnetism, usually notated by $A$, which transforms as a 1-form under coordinate changes. Since quantum fields are operator-valued, I thought it is a section of a bundle like $T^*M \otimes \mathrm{End}(\mathcal{H})$, or a complexified version of that. Here, $M$ is our spacetime, $T^*M$ is its cotangent bundle, $\mathcal{H}$ is a Hilbert space.
Soon I realized my first guess has a problem. If we consider quantum fields as sections of $(some\,geometric\,bundle)\otimes \mathrm{End}(\mathcal{H})$, it generally lacks a multiplicative structure due to the first part. For example, tensor product of $dx^{\mu}$ and $dx^\nu$ of two distinct points on $M$ would not make any sense, because they live in different cotangent fibers.
In contrast, we can define composition of two operators at two points in QFT. For example, we often compute the propagator $\langle 0| \varphi(x)\varphi(y)|0\rangle$.
Question: Quantum fields are sections of which bundles?