In December 2024, Ajay Bikmal worked on the llvm/torch-mlir repository to enhance sparse tensor workflows within MLIR, focusing on both technical depth and user experience. He enabled the Sparse Tensor Dialect across all MLIR rewrites by registering the dialect and resolving a dependency issue that previously hindered builds without StableHLO. Using C++ and leveraging his expertise in compiler design and PyTorch, Ajay also introduced and later streamlined an MPACT example to demonstrate sparsity propagation. His contributions improved build stability, clarified documentation, and expanded practical capabilities for machine learning developers working with sparse tensors in the MLIR ecosystem.