Sam Bargaoui enhanced the LanceDB repository by exposing IVF centroids through the Python API, enabling users to access exact centroid data generated during index construction without additional computation. Leveraging Rust, Python, and PyO3, Sam implemented a zero-copy interface that preserves bit-level parity between the Python layer and the underlying search routing logic. This approach eliminated the need to re-run k-means clustering in Python for analytics, reducing both compute requirements and latency. The work demonstrated a deep understanding of Arrow and FFI, resulting in richer in-database analytics and a more efficient workflow for users analyzing vector search indices.