Developed comprehensive testing coverage for Vision Transformer (ViT) models within the tenstorrent/tt-xla repository, focusing on the JAX framework. Designed and implemented a dedicated ViT tester class in Python, introducing new automated tests to validate multiple ViT configurations across various patch sizes and image resolutions. The work emphasized inference-mode coverage, laying the groundwork for future training support. By establishing a robust test harness, this effort reduced the risk of regressions and enabled safer, faster iteration on ViT features. The approach demonstrated skills in machine learning, automated testing, and transformer architectures, ensuring reliability and traceability for ongoing model development.