In Vat Photopolymerization (VPP) a part is produced by selective UV-light exposure to a photosensitive resin. This is conventionally a layer-by-layer process, although recent advances enable layerless Volumetric Additive Manufacturing (VAM). VPP is renowned for its high geometric accuracy and design freedom. These advantages have recently been exploited to replace PCB chip packaging with additively manufactured chip packages [1]. High geometric accuracy and high throughput is of key importance for this application. To achieve micron-scale accuracy precise knowledge of light transport within the material is essential to know where material will cure. Conventional modelling approaches assume rectilinear light propagation or, more commonly, they rely on volumetric light intensity kernel functions, approximations motivated primarily by low computational cost. These models fail to capture light scattering, refraction and reflection, and typically assume a one-way coupling between light transport and material curing. In reality the resin undergoes changes in absorption and refractive index during curing, which directly affects the three-dimensional light intensity field. We present an accelerated framework that accounts for the bidirectional coupling between light transport and material curing. The degree of curing, and the resulting material properties are represented as a phase-field on a finite element mesh. Light transport is modelled using a finite element compatible gradient-index (GRIN) ray tracing method [2], accurately capturing continuous refraction due to the refractive index gradients. Absorbed energy is stored on the mesh and used to drive material curing through a chemical reaction model. Results reveal optically induced geometric defects arising from the coupled opto-chemical feedback, including sawtooth profiles in VPP, and surface ripples in VAM [3]. This work provides a high-fidelity simulation tool for microscale light transport in additive manufacturing, enabling predictive modelling of part geometry and micron scale features.