Turbidity poses a major challenge for the microscopic characterization of many food systems. In these systems, local mismatches in refractive indices can cause reflection, absorption and scattering of incoming as well as outgoing light leading to significant image deterioration along sample depth. To mitigate the issue of turbidity and to increase the achievable optical resolution, we combined adaptive optics (AO) with single-molecule localization microscopy (SMLM). Building on our previously published open hardware microscopy framework, the miCube, we first added a deformable mirror to the detection path. This element enables both the compensation of aberrations directly from single-molecule data and, by further modulating the emission wavefront, the introduction of various point spread functions (PSFs) to enable SMLM in three dimensions. We further added a top hat beam shaper to the excitation path to obtain an even illumination profile across the field of view (FOV). As a model system for a non-transparent food colloid in which imaging in depth is challenging, we designed an oil-in-water emulsion in which phosvitin, a ferric ion binding protein present in from egg yolk, resides at the oil water interface. We targeted phosvitin with fluorescently labelled primary antibodies and used PSF engineering to obtain 2D and 3D images of phosvitin covered oil droplets with sub 100 nm resolution. Droplets with radii as low as 200 nm can be discerned, which is beyond the range of conventional confocal light microscopy. Our data indicated that in the model emulsion phosvitin is homogeneously distributed at the oil-water interface. With the possibility to obtain super-resolved images in depth of nontransparent colloids, our work paves the way for localizing biomacromolecules at colloidal interfaces in heterogeneous food emulsions.