Photoswitches are molecular systems that undergo reversible structural or geometrical transformations upon irradiation with light, enabling applications in molecular machines, smart materials, photopharmacology, and light-responsive materials. Photochemically triggered double-bond isomerization is widely employed in molecular photoswitches, with azobenzene representing a prominent example. Similarly, photoswitches and molecular rotors based on exocyclic C=C bond isomerization have been extensively investigated. To the best of our knowledge, photoswitches based on exocyclic C=N bond isomerization remains largely unexplored. The synthetic accessibility of exocyclic C=N bonds represent a major bottleneck, limiting the experimental realization of these motifs and the exploration of this chemical space. Recent advances in machine learning and high-throughput computational screening offer opportunities to explore large molecular libraries and identify candidates with desirable properties. However, molecular discovery alone is insufficient if the proposed structures are synthetically inaccessible. This project aims to integrate molecular design with synthetic route prediction to identify promising exocyclic C=N photoswitches that can be prepared from commercially available starting materials. Initially, a machine-learning-based retrosynthesis model, such as Synformer, will be employed to identify plausible synthetic routes for candidate molecules. The routes will be evaluated based on starting-material availability, the number of synthetic steps, predicted reaction feasibility, and expected yields. The resulting synthetically accessible candidates will define a focused chemical space for further computational investigation. The pretrained Universal Model for Atoms (UMA), a machine-learning interatomic potential, will be used to efficiently screen candidate structures based on their ground-state energies and atomic forces at substantially lower computational cost than conventional density functional theory (DFT) calculations. UMA will support conformational sampling and geometry optimization of the stable and metastable isomers. Relaxed coordinate scans and reaction-path methods will subsequently be employed to investigate possible thermal reversion mechanisms and estimate activation barriers and thermal half-lives. Candidates will be prioritized according to the predicted persistence of their metastable isomers. Selected candidates will undergo further characterization using quantum-chemical methods. Stationary-point and transition-state geometries will be re-optimized using DFT, with electronic energies refined at the DLPNO-CCSD(T)/Def2-TZVP level of theory. Intrinsic reaction coordinate (IRC) calculations will establish the connectivity between transition states and their corresponding local minima. Time-dependent DFT (TD-DFT) calculations will be used to determine the UV–Vis absorption spectra of the stable and metastable isomers, enabling assessment of their optical properties and potential for photoswitching. These results will enable the prioritization of promising candidates based on their thermal stability, energetic profiles, and optical properties, guiding the iterative refinement of the molecular library. The expected outcome is a prioritized library of promising exocyclic C=N photoswitches together with plausible synthetic routes starting from commercially available chemicals. This work will establish a computational framework that connects molecular photoswitch design with synthetic accessibility, addressing a key bottleneck in the development of new photoresponsive molecular systems. Ultimately, the project will provide a foundation for experimental synthesis and photochemical characterization of an underexplored class of molecular switches.