Recent advances in nanotechnology allow us to enhance multipolar transitions and produce multipolar photons that have much lower absorption rates than photons generated by electric dipole transitions, which are the prime choice for lasers. This property, which derives from fundamental selection rules,  will enable multipolar photons to carry and retrieve information in highly absorbing environments where standard light cannot be used.  In this project, we will model the interaction with light of emitters embedded in nanostructures with multipolar resonances, and apply machine-learning techniques for the inverse design of multipolar nanolasers and to optimize the distribution of light outside the nanolasers. The project will be done in collaboration with colleagues at the Université Côte d’Azur, France, where the students will spend some time. At the end of this thesis, the student will have acquired very sought-after skills in advanced quantum nanophotonics and machine-learning.