Worked on enhancing model loading capabilities within the tenstorrent/tt-forge-models repository, focusing on enabling end-to-end execution for a range of deep learning models. Developed dedicated PyTorch loaders for EfficientNet, Falcon, Segformer, ViT, and Vovnet, allowing seamless integration of these models into the Forge framework. Introduced abstract interfaces to standardize model decoding and sample input generation, laying the groundwork for future extensibility and simplifying the integration process. Leveraged Python and Hugging Face Transformers to ensure compatibility and maintainability. The work emphasized robust engineering practices, including traceable feature delivery, and established a solid foundation for future deep learning model support.