Contributed to the pytorch/torchchat and pytorch/executorch repositories by developing features that enhance model deployment and training workflows. In torchchat, implemented robust installation scripts using Shell and Python, improving error handling and debugging for ExecuTorch integration. Refactored the model export process in C++ and Python to embed tokenizer type information, streamlining deployment and increasing model portability. In executorch, delivered a feature enabling retrieval of method attributes within the training module, which improves metadata visibility and supports future analytics. Focused on build automation, software architecture, and unit testing, the work emphasized reliability, maintainability, and alignment with evolving project requirements.