While AlphaFold has transformed protein structure prediction, its reliance on static crystallographic data limits its ability to capture the functional plasticity of proteins. Single Particle Imaging (SPI) offers a solution by bypassing crystallisation, yet it is often hindered by low signal-to-noise ratios and radiation damage. Coulomb explosion imaging (CEI) uses particularly this radiation damage to gain structural information. I want to integrate CEI data streams into a machine learning framework, in particular Implicit Neural Representations, building on work done by Jay Shenoy (XRAI, https://github.com/JayShenoy/xrai) and previous work done by the Biophysics Research group in Uppsala, to achieve atomic-resolution in SPI. The greater goal is to enable the reconstruction of protein dynamics through ML, something Alphafold is not able to do in great detail. Ultimately, I want to integrate my models directly into the Boltz-2 (Alphafold3) architecture.