Ahmed Emam refactored the image normalization architecture in the IBM/terratorch repository, unifying the Normalize class to support both regular and temporal image data. By consolidating normalization logic and removing duplicate implementations, Ahmed centralized this functionality within the utils module, replacing the previous NormalizeWithTimesteps and streamlining code across modules. This approach improved maintainability and consistency for image processing workflows, particularly for temporal data. Working primarily in Python, Ahmed applied skills in data processing, image processing, and machine learning to enhance the repository’s extensibility. The work demonstrated thoughtful engineering depth by addressing code duplication and future-proofing normalization logic for the project.