Shunzhi Wen developed complex datatype support for tensor communication in the ProcessGroupGloo module of the pytorch/pytorch repository, enabling distributed training with complex-valued tensors on both CPU and GPU. Using C++ and Python, Shunzhi implemented a shared utilities module to handle complex data types and updated the test suite to ensure robust validation of the new feature. This work brought ProcessGroupGloo to parity with ProcessGroupNCCL, improving interoperability and deployment flexibility for teams working with complex-valued models. The depth of the implementation addressed both core communication logic and testing, reducing integration friction for distributed systems using PyTorch and complex tensors.