Worked on the SpikeInterface/spikeinterface repository to enhance neural recording preprocessing by developing a KNN-based reference channel selection feature. This approach leveraged Python and machine learning techniques to select reference channels based on a minimum number of local neighbors, improving the robustness of the preprocessing pipeline. Additionally, a backup referencing mechanism was introduced to the common_reference process, ensuring reliable operation even when local references were insufficient. The work focused on increasing the quality and reliability of neural data processing, supporting more accurate downstream analyses. Collaboration with other contributors was integral to delivering these improvements within the data processing workflow.