During May 2026, contributed a targeted optimization to the pymc-devs/pytensor repository by implementing a node rewriter that transforms solve(matrix_inverse(X), b) expressions into direct matrix multiplication using X @ b. This approach reduces reliance on explicit matrix inversion within execution graphs, streamlining the linear solve path and potentially improving both performance and numerical stability. The work involved Python programming and applied linear algebra concepts, with comprehensive unit tests added to ensure correctness across various matrix shapes and broadcasting scenarios. The focused commit lays the foundation for further mathematical operation rewrites, enhancing maintainability and clarity within the codebase.