July 2025: Delivered new tensor creation helpers for XTensorVariable in pytensor, enhancing API ergonomics and consistency with NumPy. Implemented full_like, ones_like, and zeros_like to create tensors with the same shape/dimensions as an input tensor, filled with a specified value, ones, or zeros. Added robust tests ensuring correctness and xarray compatibility, improving reliability for modeling workflows and data processing pipelines. This change reduces boilerplate, accelerates model setup, and strengthens API parity across the project.

June 2025: Delivered core tensor-manipulation capabilities for XTensorVariables, including squeeze, expand_dims, dot, and broadcast, with lowering/rewrites and comprehensive tests. Strengthened the tensor API, enabling more expressive model building and reducing the need for external utilities. Completed the feature set with robust test coverage across shapes and axes, aligning with performance and reliability goals.

May 2025 monthly summary for pymc-devs/pytensor: Delivered Transpose functionality for XTensorVariables, enabling users to reorder tensor dimensions with ellipsis notation and flexible dimension handling. Implemented a dedicated Transpose operation and supported comprehensive tests. This enhancement improves modeling flexibility and interoperability with downstream tooling, reducing boilerplate and enabling new tensor manipulation patterns.