April 2026 monthly summary for EnzymeAD projects. Focus remained on delivering distributed broadcasting capabilities, strengthening MLIR-based testing, and improving cross-repo consistency to enable scalable HPC workloads. Key features delivered: - Enzyme-JAX: Implemented MPI Bcast operation with lowering and an MLIR test, enabling data broadcast across multiple processes and paving the way for scalable parallel pipelines. - Reactant.jl: Added MPI_Bcast-based broadcast support, including the low-level MLIR operation, overlay integration, and an integration test; aligned naming (Bcast!/bcast!) for consistency across the codebase. Major bugs fixed and quality improvements: - Addressed build and formatting stability (fix clang-format) and updated operation naming to maintain consistency with existing ops. - Removed legacy tests and adjusted tests to reflect the new Bcast naming convention, improving test reliability across repos. Overall impact and accomplishments: - Enabled robust, scalable distributed data synchronization in two major repos, accelerating HPC workflows and distributed ML tasks. - Strengthened cross-repo collaboration with co-authored commits and coordinated MLIR test coverage, boosting code quality and maintainability. Technologies/skills demonstrated: - MPI-based distributed broadcasting, MLIR lowering and tests, operation revamps and naming conventions, cross-language integration (Enzyme-JAX in C++/Python? and Reactant.jl in Julia), test automation and collaboration.