Worked on enhancing type safety within the pytorch/torchrec repository by delivering a targeted feature that improved the robustness and maintainability of torch.nn.Module instances. Focused on correcting type annotations and ensuring that module methods were properly typed, the work enabled more reliable static analysis and earlier detection of type issues in continuous integration workflows. Leveraged Python and PyTorch, with an emphasis on type safety and deep learning best practices, to streamline module-level integration. These improvements reduced debugging time and eased onboarding for contributors, resulting in a more maintainable codebase and smoother development experience for machine learning practitioners using TorchRec.