Worked on integrating the MCCL backend option into the Param Comms Benchmark within the facebookresearch/param repository, focusing on backend development and performance benchmarking using Python and PyTorch. The approach involved updating supported C10d and TorchComms backends, introducing a build-time configuration flag to select MCCL, and ensuring mutual exclusivity with NCCL/NCCLX to prevent symbol conflicts. Backend validation was enhanced to recognize MCCL, enabling direct performance comparisons between MCCL and other backends within the benchmarking framework. Comprehensive documentation and commit traceability were provided, supporting data-driven vendor performance evaluations and ensuring clear integration of MCCL into the existing infrastructure.