In November 2025, Akay contributed to the sktime/sktime repository by developing an ARAR forecasting model tailored for long-memory time series data. The work involved a two-stage approach, first applying an adaptive autoregressive filter for memory shortening, followed by subset AR modeling using Yule-Walker equations. This integration enhanced the forecasting workflow’s ability to efficiently handle complex time series processes. Akay collaborated with Franz Király to document and refine the model architecture, ensuring code quality and maintainability. The project leveraged Python and data science techniques, demonstrating depth in forecasting and time series analysis while aligning with the repository’s performance objectives.