Alessandro Forina developed a data-cleaning feature for the metatensor/metatensor repository, focusing on enhancing data integrity in analytics and model training workflows. He implemented the Drop Empty Blocks operation for TensorMap, which programmatically removes blocks with zero-length dimensions to improve data quality and storage efficiency. The solution was written in Python and included comprehensive unit tests to ensure robustness across both NumPy and PyTorch backends. By addressing the challenge of noisy or redundant data, Alessandro’s work laid the foundation for future backend-agnostic tensor operations and aligned with the repository’s quality goals, demonstrating depth in cross-backend engineering and data processing.