Worked on the tenstorrent/tt-metal repository to deliver flexible tensor shape support within the Pybind module for ttnn.Tensor. The engineering approach involved removing the previous hard-coded 4D shape constraint and introducing a vector-based shape representation at the Python-C++ interface. This update, implemented using C++ and Python with a focus on bindings development, enabled dynamic tensor shapes and allowed broader integration with various models and pipelines. By addressing shape rigidity, the work improved API usability and reduced shape-related errors for downstream components, resulting in a more adaptable and robust tensor construction process within the repository’s Python bindings.