During February 2025, Nikola Vukobrat expanded the Conv2D testing framework within the tenstorrent/tt-metal repository, focusing on enhancing the TT-Forge test suite. He introduced a broad set of new test cases, including both passing and failing scenarios, to rigorously verify the correctness of the conv2d implementation. Using Python and leveraging deep learning and machine learning testing methodologies, Nikola improved regression coverage and accelerated feedback cycles in continuous integration. His work also included refining documentation and ensuring reproducibility of the testing workflow, which supports sustainable maintenance and reliable iteration on Conv2D acceleration features for future development within the project.