Developed a performance-focused optimization for co-occurrence analysis in the scverse/squidpy repository, targeting large-scale spatial datasets. The work involved re-implementing the co_occurrence function using Python and Numba, introducing a helper routine and leveraging prange-based parallel processing to accelerate computation across spatial coordinates and labels. This approach enabled higher throughput and scalability for scientific and image data analysis workflows. Documentation was updated to reflect the new implementation, and minor type-checking improvements were made to enhance maintainability. The contribution demonstrates depth in parallel computing and scientific computing, addressing a core analytics bottleneck while improving code clarity and project documentation.