Developed a custom model integration for the brain-score/vision repository, enabling seamless use of a ResNet-50 Primary Visual Cortex model within the benchmarking framework. The work involved refactoring the model implementation into a single Python file to improve maintainability and streamline onboarding. Leveraging PyTorch and Shell scripting, the integration supports loading model checkpoints directly from S3 cloud storage, enhancing reproducibility and simplifying distribution. By removing an unnecessary dependency, the solution further reduced complexity for future integrations. The model was fully registered and configured within the brainscore framework, expanding available models and strengthening the pipeline for ongoing deep learning research.