Worked on enhancing data schema merging stability in the lancedb/lance repository, focusing on sliced List and LargeList structures. Addressed a critical bug that previously caused panics during merge_with_schema operations by introducing trimmed_offsets and leveraging ListArrayExt::trimmed_offsets, ensuring compatibility with ListArray::new for sliced datasets. Validated the solution with regression tests and end-to-end checks on large datasets, confirming improved reliability and correct results. This work, implemented in Rust and utilizing Apache Arrow, reduced production risk for large or sliced datasets and strengthened the reliability of analytics pipelines, reflecting a deep understanding of backend development and data engineering challenges.