diff --git a/java/lance-jni/src/blocking_dataset.rs b/java/lance-jni/src/blocking_dataset.rs
index 203ffcdffa8..e705c0eda1d 100644
--- a/java/lance-jni/src/blocking_dataset.rs
+++ b/java/lance-jni/src/blocking_dataset.rs
@@ -38,6 +38,7 @@ use lance::dataset::{
     ColumnAlteration, CommitBuilder, Dataset, NewColumnTransform, ProjectionRequest, ReadParams,
     Version, WriteParams,
 };
+use lance::index::{DatasetIndexExt, IndexSegment};
 use lance::io::commit::namespace_manifest::LanceNamespaceExternalManifestStore;
 use lance::io::{ObjectStore, ObjectStoreParams};
 use lance::session::Session as LanceSession;
@@ -45,11 +46,10 @@ use lance::table::format::IndexMetadata;
 use lance::table::format::{BasePath, Fragment};
 use lance_core::datatypes::Schema as LanceSchema;
 use lance_file::version::LanceFileVersion;
-use lance_index::DatasetIndexExt;
 use lance_index::IndexCriteria as RustIndexCriteria;
 use lance_index::optimize::OptimizeOptions;
 use lance_index::scalar::btree::BTreeParameters;
-use lance_index::{IndexParams, IndexSegment, IndexType};
+use lance_index::{IndexParams, IndexType};
 use lance_io::object_store::ObjectStoreRegistry;
 use lance_io::object_store::StorageOptionsProvider;
 use lance_namespace::LanceNamespace;
diff --git a/python/src/dataset.rs b/python/src/dataset.rs
index 695d8b317c5..402bc010268 100644
--- a/python/src/dataset.rs
+++ b/python/src/dataset.rs
@@ -55,7 +55,7 @@ use lance::dataset::{
     transaction::{Operation, Transaction},
 };
 use lance::index::vector::utils::get_vector_type;
-use lance::index::{DatasetIndexInternalExt, vector::VectorIndexParams};
+use lance::index::{DatasetIndexExt, DatasetIndexInternalExt, vector::VectorIndexParams};
 use lance::{dataset::builder::DatasetBuilder, index::vector::IndexFileVersion};
 use lance_arrow::as_fixed_size_list_array;
 use lance_core::Error;
@@ -67,7 +67,7 @@ use lance_index::scalar::inverted::query::{
     BooleanQuery, BoostQuery, FtsQuery, MatchQuery, MultiMatchQuery, Operator, PhraseQuery,
 };
 use lance_index::{
-    DatasetIndexExt, IndexParams, IndexType,
+    IndexParams, IndexType,
     optimize::OptimizeOptions,
     scalar::{FullTextSearchQuery, InvertedIndexParams, ScalarIndexParams},
     vector::{
diff --git a/python/src/indices.rs b/python/src/indices.rs
index f1d42918962..9651c6cc00e 100644
--- a/python/src/indices.rs
+++ b/python/src/indices.rs
@@ -11,6 +11,7 @@ use arrow_data::ArrayData;
 use chrono::{DateTime, Utc};
 use lance::dataset::Dataset as LanceDataset;
 use lance::index::vector::ivf::builder::write_vector_storage;
+use lance::index::{DatasetIndexExt, IndexSegment, IndexSegmentPlan};
 use lance::io::ObjectStore;
 use lance_index::progress::NoopIndexBuildProgress;
 use lance_index::vector::ivf::shuffler::{IvfShuffler, shuffle_vectors};
@@ -37,7 +38,7 @@ use crate::{
     dataset::Dataset, error::PythonErrorExt, file::object_store_from_uri_or_path_no_options, rt,
 };
 use lance::index::vector::ivf::write_ivf_pq_file_from_existing_index;
-use lance_index::{DatasetIndexExt, IndexDescription, IndexSegment, IndexSegmentPlan, IndexType};
+use lance_index::{IndexDescription, IndexType};
 use uuid::Uuid;
 
 #[pyclass(name = "IndexConfig", module = "lance.indices", get_all)]
diff --git a/python/src/lib.rs b/python/src/lib.rs
index f7d233cf117..9730f2ba1c5 100644
--- a/python/src/lib.rs
+++ b/python/src/lib.rs
@@ -34,6 +34,7 @@ use ::arrow::pyarrow::PyArrowType;
 use ::arrow_schema::Schema as ArrowSchema;
 use ::lance::arrow::json::ArrowJsonExt;
 use ::lance::datafusion::LanceTableProvider;
+use ::lance::index::DatasetIndexExt;
 use datafusion_ffi::proto::logical_extension_codec::FFI_LogicalExtensionCodec;
 use datafusion_ffi::table_provider::FFI_TableProvider;
 #[cfg(feature = "datagen")]
@@ -52,7 +53,6 @@ use file::{
     LanceBufferDescriptor, LanceColumnMetadata, LanceFileMetadata, LanceFileReader,
     LanceFileStatistics, LanceFileWriter, LancePageMetadata, stable_version,
 };
-use lance_index::DatasetIndexExt;
 use log::Level;
 use pyo3::exceptions::PyIOError;
 use pyo3::prelude::*;
diff --git a/rust/examples/src/full_text_search.rs b/rust/examples/src/full_text_search.rs
index 22e701c5863..8269f590ee8 100644
--- a/rust/examples/src/full_text_search.rs
+++ b/rust/examples/src/full_text_search.rs
@@ -15,8 +15,8 @@ use arrow::datatypes::UInt64Type;
 use arrow_schema::{DataType, Field, Schema};
 use itertools::Itertools;
 use lance::Dataset;
+use lance::index::DatasetIndexExt;
 use lance_datagen::{RowCount, array};
-use lance_index::DatasetIndexExt;
 use lance_index::scalar::inverted::flat_full_text_search;
 use lance_index::scalar::{FullTextSearchQuery, InvertedIndexParams};
 use object_store::path::Path;
diff --git a/rust/examples/src/ivf_hnsw.rs b/rust/examples/src/ivf_hnsw.rs
index 296c03b19d6..c1898e10682 100644
--- a/rust/examples/src/ivf_hnsw.rs
+++ b/rust/examples/src/ivf_hnsw.rs
@@ -11,11 +11,12 @@ use clap::Parser;
 use futures::TryStreamExt;
 use lance::Dataset;
 use lance::dataset::ProjectionRequest;
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
+use lance_index::IndexType;
 use lance_index::vector::hnsw::builder::HnswBuildParams;
 use lance_index::vector::ivf::IvfBuildParams;
 use lance_index::vector::sq::builder::SQBuildParams;
-use lance_index::{DatasetIndexExt, IndexType};
 use lance_linalg::distance::MetricType;
 
 #[derive(Parser, Debug)]
diff --git a/rust/lance-index/src/lib.rs b/rust/lance-index/src/lib.rs
index bb00a716173..62ae68414a6 100644
--- a/rust/lance-index/src/lib.rs
+++ b/rust/lance-index/src/lib.rs
@@ -29,11 +29,9 @@ pub mod progress;
 pub mod registry;
 pub mod scalar;
 pub mod traits;
-pub mod types;
 pub mod vector;
 
 pub use crate::traits::*;
-pub use crate::types::{IndexSegment, IndexSegmentPlan};
 
 pub const INDEX_FILE_NAME: &str = "index.idx";
 /// The name of the auxiliary index file.
diff --git a/rust/lance-index/src/traits.rs b/rust/lance-index/src/traits.rs
index a883c3ce494..130e59cad81 100644
--- a/rust/lance-index/src/traits.rs
+++ b/rust/lance-index/src/traits.rs
@@ -1,13 +1,8 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The Lance Authors
 
-use std::sync::Arc;
+use lance_core::Result;
 
-use async_trait::async_trait;
-use datafusion::execution::SendableRecordBatchStream;
-use lance_core::{Error, Result};
-
-use crate::{IndexParams, IndexType, optimize::OptimizeOptions, types::IndexSegment};
 use lance_table::format::IndexMetadata;
 
 /// A set of criteria used to filter potential indices to use for a query
@@ -121,191 +116,3 @@ pub trait IndexDescription: Send + Sync {
     /// (for backward compatibility with indices created before file tracking was added).
     fn total_size_bytes(&self) -> Option<u64>;
 }
-
-// Extends Lance Dataset with secondary index.
-#[async_trait]
-pub trait DatasetIndexExt {
-    type IndexBuilder<'a>
-    where
-        Self: 'a;
-    type IndexSegmentBuilder<'a>
-    where
-        Self: 'a;
-
-    /// Create a builder for creating an index on columns.
-    ///
-    /// This returns a builder that can be configured with additional options
-    /// like `name()`, `replace()`, and `train()` before awaiting to execute.
-    ///
-    /// # Parameters
-    /// - `columns`: the columns to build the indices on.
-    /// - `index_type`: specify [`IndexType`].
-    /// - `params`: index parameters.
-    fn create_index_builder<'a>(
-        &'a mut self,
-        columns: &'a [&'a str],
-        index_type: IndexType,
-        params: &'a dyn IndexParams,
-    ) -> Self::IndexBuilder<'a>;
-
-    /// Create a builder for building physical index segments from uncommitted
-    /// vector index outputs.
-    ///
-    /// The caller supplies the uncommitted index metadata returned by
-    /// `execute_uncommitted()` so the builder can plan segment grouping without
-    /// rediscovering fragment coverage.
-    ///
-    /// This is the canonical entry point for distributed vector segment build.
-    /// After building the physical segments, publish them as a
-    /// logical index with [`Self::commit_existing_index_segments`].
-    fn create_index_segment_builder<'a>(&'a self) -> Self::IndexSegmentBuilder<'a>;
-
-    /// Create indices on columns.
-    ///
-    /// Upon finish, a new dataset version is generated.
-    ///
-    /// Parameters:
-    ///
-    ///  - `columns`: the columns to build the indices on.
-    ///  - `index_type`: specify [`IndexType`].
-    ///  - `name`: optional index name. Must be unique in the dataset.
-    ///            if not provided, it will auto-generate one.
-    ///  - `params`: index parameters.
-    ///  - `replace`: replace the existing index if it exists.
-    ///
-    /// Returns the metadata of the created index.
-    async fn create_index(
-        &mut self,
-        columns: &[&str],
-        index_type: IndexType,
-        name: Option<String>,
-        params: &dyn IndexParams,
-        replace: bool,
-    ) -> Result<IndexMetadata>;
-
-    /// Drop indices by name.
-    ///
-    /// Upon finish, a new dataset version is generated.
-    ///
-    /// Parameters:
-    ///
-    /// - `name`: the name of the index to drop.
-    async fn drop_index(&mut self, name: &str) -> Result<()>;
-
-    /// Prewarm an index by name.
-    ///
-    /// This will load the index into memory and cache it.
-    ///
-    /// Generally, this should only be called when it is known the entire index will
-    /// fit into the index cache.
-    ///
-    /// This is a hint that is not enforced by all indices today.  Some indices may choose
-    /// to ignore this hint.
-    async fn prewarm_index(&self, name: &str) -> Result<()>;
-
-    /// Read all indices of this Dataset version.
-    ///
-    /// The indices are lazy loaded and cached in memory within the `Dataset` instance.
-    /// The cache is invalidated when the dataset version (Manifest) is changed.
-    async fn load_indices(&self) -> Result<Arc<Vec<IndexMetadata>>>;
-
-    /// Loads all the indies of a given UUID.
-    ///
-    /// Note that it is possible to have multiple indices with the same UUID,
-    /// as they are the deltas of the same index.
-    async fn load_index(&self, uuid: &str) -> Result<Option<IndexMetadata>> {
-        self.load_indices().await.map(|indices| {
-            indices
-                .iter()
-                .find(|idx| idx.uuid.to_string() == uuid)
-                .cloned()
-        })
-    }
-
-    /// Loads a specific index with the given index name
-    ///
-    /// Returns
-    /// -------
-    /// - `Ok(indices)`: if the index exists, returns the index.
-    /// - `Ok(vec![])`: if the index does not exist.
-    /// - `Err(e)`: if there is an error loading indices.
-    ///
-    async fn load_indices_by_name(&self, name: &str) -> Result<Vec<IndexMetadata>> {
-        self.load_indices().await.map(|indices| {
-            indices
-                .iter()
-                .filter(|idx| idx.name == name)
-                .cloned()
-                .collect()
-        })
-    }
-
-    /// Loads a specific index with the given index name.
-    /// This function only works for indices that are unique.
-    /// If there are multiple indices sharing the same name, please use [`Self::load_indices_by_name`]
-    ///
-    /// Returns
-    /// -------
-    /// - `Ok(Some(index))`: if the index exists, returns the index.
-    /// - `Ok(None)`: if the index does not exist.
-    /// - `Err(e)`: Index error if there are multiple indexes sharing the same name.
-    ///
-    async fn load_index_by_name(&self, name: &str) -> Result<Option<IndexMetadata>> {
-        let indices = self.load_indices_by_name(name).await?;
-        if indices.is_empty() {
-            Ok(None)
-        } else if indices.len() == 1 {
-            Ok(Some(indices[0].clone()))
-        } else {
-            Err(Error::index(format!(
-                "Found multiple indices of the same name: {:?}, please use load_indices_by_name",
-                indices.iter().map(|idx| &idx.name).collect::<Vec<_>>()
-            )))
-        }
-    }
-
-    /// Describes indexes in a dataset
-    ///
-    /// This method should only access the index metadata and should not load the index into memory.
-    ///
-    /// More detailed information may be available from `index_statistics` but that will require
-    /// loading the index into memory.
-    async fn describe_indices<'a, 'b>(
-        &'a self,
-        criteria: Option<IndexCriteria<'b>>,
-    ) -> Result<Vec<Arc<dyn IndexDescription>>>;
-
-    /// Loads a specific index with the given index name.
-    async fn load_scalar_index<'a, 'b>(
-        &'a self,
-        criteria: IndexCriteria<'b>,
-    ) -> Result<Option<IndexMetadata>>;
-
-    /// Optimize indices.
-    async fn optimize_indices(&mut self, options: &OptimizeOptions) -> Result<()>;
-
-    /// Find index with a given index_name and return its serialized statistics.
-    ///
-    /// If the index does not exist, return Error.
-    async fn index_statistics(&self, index_name: &str) -> Result<String>;
-
-    /// Commit one or more existing physical index segments as a logical index.
-    ///
-    /// This publishes already-built physical segments. It does not build
-    /// or merge index data; callers should first build segments with
-    /// [`Self::create_index_segment_builder`] or another index-specific build
-    /// path and then pass the resulting segments here.
-    async fn commit_existing_index_segments(
-        &mut self,
-        index_name: &str,
-        column: &str,
-        segments: Vec<IndexSegment>,
-    ) -> Result<()>;
-
-    async fn read_index_partition(
-        &self,
-        index_name: &str,
-        partition_id: usize,
-        with_vector: bool,
-    ) -> Result<SendableRecordBatchStream>;
-}
diff --git a/rust/lance-index/src/types.rs b/rust/lance-index/src/types.rs
deleted file mode 100644
index 4d653f21f13..00000000000
--- a/rust/lance-index/src/types.rs
+++ /dev/null
@@ -1,124 +0,0 @@
-// SPDX-License-Identifier: Apache-2.0
-// SPDX-FileCopyrightText: Copyright The Lance Authors
-
-use std::sync::Arc;
-
-use crate::IndexType;
-use lance_table::format::IndexMetadata;
-use roaring::RoaringBitmap;
-use uuid::Uuid;
-
-/// A single physical segment of a logical index.
-///
-/// Each segment is stored independently and will become one manifest entry when committed.
-/// The logical index identity (name / target column / dataset version) is provided separately
-/// by the commit API.
-#[derive(Debug, Clone, PartialEq)]
-pub struct IndexSegment {
-    /// Unique ID of the physical segment.
-    uuid: Uuid,
-    /// The fragments covered by this segment.
-    fragment_bitmap: RoaringBitmap,
-    /// Metadata specific to the index type.
-    index_details: Arc<prost_types::Any>,
-    /// The on-disk index version for this segment.
-    index_version: i32,
-}
-
-impl IndexSegment {
-    /// Create a fully described segment with the given UUID, fragment coverage, and index
-    /// metadata.
-    pub fn new<I>(
-        uuid: Uuid,
-        fragment_bitmap: I,
-        index_details: Arc<prost_types::Any>,
-        index_version: i32,
-    ) -> Self
-    where
-        I: IntoIterator<Item = u32>,
-    {
-        Self {
-            uuid,
-            fragment_bitmap: fragment_bitmap.into_iter().collect(),
-            index_details,
-            index_version,
-        }
-    }
-
-    /// Return the UUID of this segment.
-    pub fn uuid(&self) -> Uuid {
-        self.uuid
-    }
-
-    /// Return the fragment coverage of this segment.
-    pub fn fragment_bitmap(&self) -> &RoaringBitmap {
-        &self.fragment_bitmap
-    }
-
-    /// Return the serialized index details for this segment.
-    pub fn index_details(&self) -> &Arc<prost_types::Any> {
-        &self.index_details
-    }
-
-    /// Return the on-disk index version for this segment.
-    pub fn index_version(&self) -> i32 {
-        self.index_version
-    }
-
-    /// Consume the segment and return its component parts.
-    pub fn into_parts(self) -> (Uuid, RoaringBitmap, Arc<prost_types::Any>, i32) {
-        (
-            self.uuid,
-            self.fragment_bitmap,
-            self.index_details,
-            self.index_version,
-        )
-    }
-}
-
-/// A plan for building one physical segment from one or more existing
-/// vector index segments.
-#[derive(Debug, Clone, PartialEq)]
-pub struct IndexSegmentPlan {
-    segment: IndexSegment,
-    segments: Vec<IndexMetadata>,
-    estimated_bytes: u64,
-    requested_index_type: Option<IndexType>,
-}
-
-impl IndexSegmentPlan {
-    /// Create a plan for one built segment.
-    pub fn new(
-        segment: IndexSegment,
-        segments: Vec<IndexMetadata>,
-        estimated_bytes: u64,
-        requested_index_type: Option<IndexType>,
-    ) -> Self {
-        Self {
-            segment,
-            segments,
-            estimated_bytes,
-            requested_index_type,
-        }
-    }
-
-    /// Return the segment metadata that should be committed after this plan is built.
-    pub fn segment(&self) -> &IndexSegment {
-        &self.segment
-    }
-
-    /// Return the input segment metadata that should be combined into the segment.
-    pub fn segments(&self) -> &[IndexMetadata] {
-        &self.segments
-    }
-
-    /// Return the estimated number of bytes covered by this plan.
-    pub fn estimated_bytes(&self) -> u64 {
-        self.estimated_bytes
-    }
-
-    /// Return the requested logical index type, if one was supplied to the planner.
-    pub fn requested_index_type(&self) -> Option<IndexType> {
-        self.requested_index_type
-    }
-}
diff --git a/rust/lance-namespace-impls/src/dir/manifest.rs b/rust/lance-namespace-impls/src/dir/manifest.rs
index 902c5462acc..9c8bf337712 100644
--- a/rust/lance-namespace-impls/src/dir/manifest.rs
+++ b/rust/lance-namespace-impls/src/dir/manifest.rs
@@ -18,6 +18,7 @@ use lance::dataset::{
     DeleteBuilder, MergeInsertBuilder, ReadParams, WhenMatched, WhenNotMatched, WriteParams,
     builder::DatasetBuilder,
 };
+use lance::index::DatasetIndexExt;
 use lance::session::Session;
 use lance::{Dataset, dataset::scanner::Scanner};
 use lance_core::Error as LanceError;
@@ -26,7 +27,6 @@ use lance_core::{Error, Result, box_error};
 use lance_index::IndexType;
 use lance_index::optimize::OptimizeOptions;
 use lance_index::scalar::{BuiltinIndexType, ScalarIndexParams};
-use lance_index::traits::DatasetIndexExt;
 use lance_io::object_store::{ObjectStore, ObjectStoreParams};
 use lance_namespace::LanceNamespace;
 use lance_namespace::error::NamespaceError;
diff --git a/rust/lance/benches/ivf_pq.rs b/rust/lance/benches/ivf_pq.rs
index ebe70a32d4d..ae92b406168 100644
--- a/rust/lance/benches/ivf_pq.rs
+++ b/rust/lance/benches/ivf_pq.rs
@@ -9,13 +9,14 @@ use arrow_array::{FixedSizeListArray, RecordBatch, RecordBatchIterator};
 use arrow_schema::{DataType, Field, FieldRef, Schema};
 use criterion::{Criterion, criterion_group, criterion_main};
 
+use lance::index::DatasetIndexExt;
 use lance::{
     Dataset,
     dataset::{WriteMode, WriteParams},
     index::vector::VectorIndexParams,
 };
 use lance_arrow::*;
-use lance_index::{DatasetIndexExt, IndexType};
+use lance_index::IndexType;
 use lance_linalg::distance::MetricType;
 use lance_testing::datagen::generate_random_array;
 #[cfg(target_os = "linux")]
diff --git a/rust/lance/benches/mem_wal_write.rs b/rust/lance/benches/mem_wal_write.rs
index 363767b519e..31f855fd0ad 100644
--- a/rust/lance/benches/mem_wal_write.rs
+++ b/rust/lance/benches/mem_wal_write.rs
@@ -54,12 +54,13 @@ use arrow_schema::{DataType, Field, Schema as ArrowSchema};
 use criterion::{BenchmarkId, Criterion, Throughput, criterion_group, criterion_main};
 use lance::dataset::mem_wal::{DatasetMemWalExt, MemWalConfig, RegionWriterConfig};
 use lance::dataset::{Dataset, WriteParams};
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
 use lance_arrow::FixedSizeListArrayExt;
+use lance_index::IndexType;
 use lance_index::scalar::ScalarIndexParams;
 use lance_index::vector::ivf::IvfBuildParams;
 use lance_index::vector::pq::PQBuildParams;
-use lance_index::{DatasetIndexExt, IndexType};
 use lance_linalg::distance::DistanceType;
 #[cfg(target_os = "linux")]
 use pprof::criterion::{Output, PProfProfiler};
diff --git a/rust/lance/benches/memtable_read.rs b/rust/lance/benches/memtable_read.rs
index 240c90294e9..efd16dbb62a 100644
--- a/rust/lance/benches/memtable_read.rs
+++ b/rust/lance/benches/memtable_read.rs
@@ -38,15 +38,16 @@ use criterion::{BenchmarkId, Criterion, Throughput, criterion_group, criterion_m
 use futures::TryStreamExt;
 use lance::dataset::mem_wal::write::{CacheConfig, IndexStore, MemTable};
 use lance::dataset::{Dataset, WriteParams};
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
 use lance_arrow::FixedSizeListArrayExt;
+use lance_index::IndexType;
 use lance_index::scalar::FullTextSearchQuery;
 use lance_index::scalar::inverted::tokenizer::InvertedIndexParams;
 use lance_index::vector::ivf::IvfBuildParams;
 use lance_index::vector::ivf::storage::IvfModel;
 use lance_index::vector::kmeans::{KMeansParams, train_kmeans};
 use lance_index::vector::pq::builder::PQBuildParams;
-use lance_index::{DatasetIndexExt, IndexType};
 use lance_linalg::distance::{DistanceType, MetricType};
 #[cfg(target_os = "linux")]
 use pprof::criterion::{Output, PProfProfiler};
diff --git a/rust/lance/benches/vector_index.rs b/rust/lance/benches/vector_index.rs
index cf4106db8b4..21c9aa4e4aa 100644
--- a/rust/lance/benches/vector_index.rs
+++ b/rust/lance/benches/vector_index.rs
@@ -15,10 +15,11 @@ use pprof::criterion::{Output, PProfProfiler};
 use rand::Rng;
 
 use lance::dataset::{Dataset, WriteMode, WriteParams, builder::DatasetBuilder};
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
 use lance_arrow::{FixedSizeListArrayExt, as_fixed_size_list_array};
 use lance_index::{
-    DatasetIndexExt, IndexType,
+    IndexType,
     vector::{ivf::IvfBuildParams, pq::PQBuildParams},
 };
 use lance_linalg::distance::MetricType;
diff --git a/rust/lance/benches/vector_throughput.rs b/rust/lance/benches/vector_throughput.rs
index ff24c7771ba..9a04971684b 100644
--- a/rust/lance/benches/vector_throughput.rs
+++ b/rust/lance/benches/vector_throughput.rs
@@ -19,10 +19,11 @@ use pprof::criterion::{Output, PProfProfiler};
 use rand::Rng;
 
 use lance::dataset::{Dataset, WriteMode, WriteParams};
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
 use lance_arrow::FixedSizeListArrayExt;
 use lance_index::{
-    DatasetIndexExt, IndexType,
+    IndexType,
     vector::{ivf::IvfBuildParams, pq::PQBuildParams},
 };
 use lance_linalg::distance::MetricType;
diff --git a/rust/lance/src/bin/lq.rs b/rust/lance/src/bin/lq.rs
index d28121bbf55..afaa15c7f10 100644
--- a/rust/lance/src/bin/lq.rs
+++ b/rust/lance/src/bin/lq.rs
@@ -10,9 +10,9 @@ use futures::TryStreamExt;
 use futures::stream::StreamExt;
 
 use lance::dataset::Dataset;
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
 use lance::{Error, Result};
-use lance_index::DatasetIndexExt;
 use lance_linalg::distance::MetricType;
 
 #[derive(Parser)]
diff --git a/rust/lance/src/dataset.rs b/rust/lance/src/dataset.rs
index 02f4b28e047..8f3deb2bef6 100644
--- a/rust/lance/src/dataset.rs
+++ b/rust/lance/src/dataset.rs
@@ -15,6 +15,7 @@ use futures::{FutureExt, Stream};
 
 use crate::dataset::metadata::UpdateFieldMetadataBuilder;
 use crate::dataset::transaction::translate_schema_metadata_updates;
+use crate::index::DatasetIndexExt;
 use crate::session::caches::{DSMetadataCache, ManifestKey, TransactionKey};
 use crate::session::index_caches::DSIndexCache;
 use itertools::Itertools;
@@ -30,7 +31,7 @@ use lance_datafusion::projection::ProjectionPlan;
 use lance_file::datatypes::populate_schema_dictionary;
 use lance_file::reader::FileReaderOptions;
 use lance_file::version::LanceFileVersion;
-use lance_index::{DatasetIndexExt, IndexType};
+use lance_index::IndexType;
 use lance_io::object_store::{
     LanceNamespaceStorageOptionsProvider, ObjectStore, ObjectStoreParams, StorageOptions,
     StorageOptionsAccessor, StorageOptionsProvider,
diff --git a/rust/lance/src/dataset/cleanup.rs b/rust/lance/src/dataset/cleanup.rs
index 18380458eeb..f5162bba191 100644
--- a/rust/lance/src/dataset/cleanup.rs
+++ b/rust/lance/src/dataset/cleanup.rs
@@ -1180,6 +1180,7 @@ mod tests {
 
     use super::*;
     use crate::blob::{BlobArrayBuilder, blob_field};
+    use crate::index::DatasetIndexExt;
     use crate::{
         dataset::{ReadParams, WriteMode, WriteParams, builder::DatasetBuilder},
         index::vector::VectorIndexParams,
@@ -1193,7 +1194,7 @@ mod tests {
     use datafusion::common::assert_contains;
     use lance_core::utils::tempfile::TempStrDir;
     use lance_core::utils::testing::{ProxyObjectStore, ProxyObjectStorePolicy};
-    use lance_index::{DatasetIndexExt, IndexType};
+    use lance_index::IndexType;
     use lance_io::object_store::{
         ObjectStore, ObjectStoreParams, ObjectStoreRegistry, WrappingObjectStore,
     };
@@ -2642,8 +2643,9 @@ mod tests {
         // Create a full-text index (Inverted) on the "text" column once.
         // We only create this on main during dataset creation. Branches inherit the index configuration.
         async fn create_text_index(&mut self) -> Result<()> {
+            use crate::index::DatasetIndexExt;
+            use lance_index::IndexType;
             use lance_index::scalar::InvertedIndexParams;
-            use lance_index::{DatasetIndexExt, IndexType};
             let params = InvertedIndexParams::default();
             self.dataset
                 .create_index(&["text"], IndexType::Inverted, None, &params, true)
diff --git a/rust/lance/src/dataset/index.rs b/rust/lance/src/dataset/index.rs
index d4de4d0be6f..91c535468e5 100644
--- a/rust/lance/src/dataset/index.rs
+++ b/rust/lance/src/dataset/index.rs
@@ -9,13 +9,13 @@ use std::sync::Arc;
 use crate::Dataset;
 use crate::dataset::optimize::RemappedIndex;
 use crate::dataset::optimize::remapping::RemapResult;
+use crate::index::DatasetIndexExt;
 use crate::index::remap_index;
 use crate::index::scalar::infer_scalar_index_details;
 use arrow_schema::DataType;
 use async_trait::async_trait;
 use lance_core::{Error, Result};
 use lance_encoding::version::LanceFileVersion;
-use lance_index::DatasetIndexExt;
 use lance_index::frag_reuse::FRAG_REUSE_INDEX_NAME;
 use lance_index::scalar::lance_format::LanceIndexStore;
 use lance_table::format::IndexMetadata;
diff --git a/rust/lance/src/dataset/index/frag_reuse.rs b/rust/lance/src/dataset/index/frag_reuse.rs
index cc7b845313d..4fbefcd4725 100644
--- a/rust/lance/src/dataset/index/frag_reuse.rs
+++ b/rust/lance/src/dataset/index/frag_reuse.rs
@@ -149,12 +149,13 @@ fn is_index_remap_caught_up(
 mod tests {
     use super::*;
     use crate::dataset::optimize::{CompactionOptions, compact_files, remapping};
+    use crate::index::DatasetIndexExt;
     use crate::utils::test::{DatagenExt, FragmentCount, FragmentRowCount};
     use all_asserts::{assert_false, assert_true};
     use arrow_array::types::{Float32Type, Int32Type};
     use lance_datagen::Dimension;
+    use lance_index::IndexType;
     use lance_index::scalar::ScalarIndexParams;
-    use lance_index::{DatasetIndexExt, IndexType};
 
     #[tokio::test]
     async fn test_cleanup_frag_reuse_index() {
diff --git a/rust/lance/src/dataset/mem_wal/api.rs b/rust/lance/src/dataset/mem_wal/api.rs
index 99bb415d6a0..30c6a10811a 100644
--- a/rust/lance/src/dataset/mem_wal/api.rs
+++ b/rust/lance/src/dataset/mem_wal/api.rs
@@ -8,9 +8,9 @@
 
 use std::sync::Arc;
 
+use crate::index::DatasetIndexExt;
 use async_trait::async_trait;
 use lance_core::{Error, Result};
-use lance_index::DatasetIndexExt;
 use lance_index::mem_wal::{MEM_WAL_INDEX_NAME, MemWalIndexDetails, RegionSpec};
 use lance_index::vector::ivf::storage::IvfModel;
 use lance_index::vector::pq::ProductQuantizer;
diff --git a/rust/lance/src/dataset/mem_wal/memtable/flush.rs b/rust/lance/src/dataset/mem_wal/memtable/flush.rs
index c3b955fc265..9ff133413ba 100644
--- a/rust/lance/src/dataset/mem_wal/memtable/flush.rs
+++ b/rust/lance/src/dataset/mem_wal/memtable/flush.rs
@@ -962,7 +962,7 @@ mod tests {
     #[tokio::test]
     async fn test_flusher_with_btree_index() {
         use super::super::super::index::{BTreeIndexConfig, IndexStore};
-        use lance_index::DatasetIndexExt;
+        use crate::index::DatasetIndexExt;
 
         let (store, base_path, base_uri, _temp_dir) = create_local_store().await;
         let region_id = Uuid::new_v4();
@@ -1060,9 +1060,9 @@ mod tests {
     #[tokio::test]
     async fn test_flusher_with_ivf_pq_index() {
         use super::super::super::index::{IndexStore, IvfPqIndexConfig};
+        use crate::index::DatasetIndexExt;
         use arrow_array::{FixedSizeListArray, Float32Array};
         use lance_arrow::FixedSizeListArrayExt;
-        use lance_index::DatasetIndexExt;
         use lance_index::vector::ivf::storage::IvfModel;
         use lance_index::vector::kmeans::{KMeansParams, train_kmeans};
         use lance_index::vector::pq::PQBuildParams;
@@ -1284,9 +1284,9 @@ mod tests {
     #[tokio::test]
     async fn test_flusher_with_fts_index() {
         use super::super::super::index::{FtsIndexConfig, IndexStore};
+        use crate::index::DatasetIndexExt;
         use arrow_array::StringArray;
         use arrow_schema::{DataType, Field, Schema as ArrowSchema};
-        use lance_index::DatasetIndexExt;
         use std::sync::Arc;
 
         let (store, base_path, base_uri, _temp_dir) = create_local_store().await;
diff --git a/rust/lance/src/dataset/mem_wal/write.rs b/rust/lance/src/dataset/mem_wal/write.rs
index 65aa1d97586..dda06a7c9d9 100644
--- a/rust/lance/src/dataset/mem_wal/write.rs
+++ b/rust/lance/src/dataset/mem_wal/write.rs
@@ -2128,16 +2128,17 @@ mod tests {
 mod region_writer_tests {
     use std::sync::Arc;
 
+    use crate::index::DatasetIndexExt;
     use arrow_array::{
         FixedSizeListArray, Float32Array, Int64Array, RecordBatch, RecordBatchIterator, StringArray,
     };
     use arrow_schema::{DataType, Field, Schema as ArrowSchema};
     use lance_arrow::FixedSizeListArrayExt;
+    use lance_index::IndexType;
     use lance_index::scalar::ScalarIndexParams;
     use lance_index::scalar::inverted::InvertedIndexParams;
     use lance_index::vector::ivf::IvfBuildParams;
     use lance_index::vector::pq::builder::PQBuildParams;
-    use lance_index::{DatasetIndexExt, IndexType};
     use lance_linalg::distance::MetricType;
     use uuid::Uuid;
 
diff --git a/rust/lance/src/dataset/optimize.rs b/rust/lance/src/dataset/optimize.rs
index 7cec534e171..c70eb93bbcd 100644
--- a/rust/lance/src/dataset/optimize.rs
+++ b/rust/lance/src/dataset/optimize.rs
@@ -97,6 +97,7 @@ use super::{WriteMode, WriteParams, write_fragments_internal};
 use crate::Dataset;
 use crate::Result;
 use crate::dataset::utils::CapturedRowIds;
+use crate::index::DatasetIndexExt;
 use crate::io::commit::{commit_transaction, migrate_fragments};
 use datafusion::physical_plan::SendableRecordBatchStream;
 use datafusion::physical_plan::stream::RecordBatchStreamAdapter;
@@ -105,7 +106,6 @@ use lance_core::Error;
 use lance_core::datatypes::BlobHandling;
 use lance_core::utils::tokio::get_num_compute_intensive_cpus;
 use lance_core::utils::tracing::{DATASET_COMPACTING_EVENT, TRACE_DATASET_EVENTS};
-use lance_index::DatasetIndexExt;
 use lance_index::frag_reuse::FragReuseGroup;
 use lance_table::format::{Fragment, RowIdMeta};
 use roaring::{RoaringBitmap, RoaringTreemap};
diff --git a/rust/lance/src/dataset/optimize/remapping.rs b/rust/lance/src/dataset/optimize/remapping.rs
index 253bd63ff15..dab62bf6166 100644
--- a/rust/lance/src/dataset/optimize/remapping.rs
+++ b/rust/lance/src/dataset/optimize/remapping.rs
@@ -6,12 +6,12 @@
 
 use crate::Result;
 use crate::dataset::transaction::{Operation, Transaction};
+use crate::index::DatasetIndexExt;
 use crate::index::frag_reuse::{load_frag_reuse_index_details, open_frag_reuse_index};
 use crate::{Dataset, index};
 use async_trait::async_trait;
 use lance_core::Error;
 use lance_core::utils::address::RowAddress;
-use lance_index::DatasetIndexExt;
 use lance_index::frag_reuse::{FRAG_REUSE_INDEX_NAME, FragDigest};
 use lance_table::format::{Fragment, IndexFile, IndexMetadata};
 use lance_table::io::manifest::read_manifest_indexes;
diff --git a/rust/lance/src/dataset/rowids.rs b/rust/lance/src/dataset/rowids.rs
index b6b4b106798..447868fca99 100644
--- a/rust/lance/src/dataset/rowids.rs
+++ b/rust/lance/src/dataset/rowids.rs
@@ -130,6 +130,7 @@ mod test {
     use super::*;
 
     use crate::dataset::optimize::{CompactionOptions, compact_files};
+    use crate::index::DatasetIndexExt;
     use crate::utils::test::{DatagenExt, FragmentCount, FragmentRowCount};
     use arrow_array::cast::AsArray;
     use arrow_array::types::{Float32Type, Int32Type, UInt64Type};
@@ -139,7 +140,7 @@ mod test {
     use lance_core::datatypes::Schema;
     use lance_core::{ROW_ADDR, ROW_ID, utils::address::RowAddress};
     use lance_datagen::Dimension;
-    use lance_index::{DatasetIndexExt, IndexType, scalar::ScalarIndexParams};
+    use lance_index::{IndexType, scalar::ScalarIndexParams};
     use std::collections::HashMap;
     use std::collections::HashSet;
 
diff --git a/rust/lance/src/dataset/scanner.rs b/rust/lance/src/dataset/scanner.rs
index 63bd7884879..34483bfd847 100644
--- a/rust/lance/src/dataset/scanner.rs
+++ b/rust/lance/src/dataset/scanner.rs
@@ -6,6 +6,7 @@ use std::pin::Pin;
 use std::sync::{Arc, LazyLock};
 use std::task::{Context, Poll};
 
+use crate::index::DatasetIndexExt;
 use arrow::array::AsArray;
 use arrow_array::{Array, Float32Array, Int64Array, RecordBatch};
 use arrow_schema::{DataType, Field as ArrowField, Schema as ArrowSchema, SchemaRef, SortOptions};
@@ -60,13 +61,13 @@ use lance_datafusion::projection::ProjectionPlan;
 use lance_file::reader::FileReaderOptions;
 use lance_index::IndexCriteria;
 use lance_index::scalar::FullTextSearchQuery;
+use lance_index::scalar::expression::ScalarIndexExpr;
 use lance_index::scalar::expression::{INDEX_EXPR_RESULT_SCHEMA, IndexExprResult, PlannerIndexExt};
 use lance_index::scalar::inverted::query::{
     FtsQuery, FtsQueryNode, FtsSearchParams, MatchQuery, PhraseQuery, fill_fts_query_column,
 };
 use lance_index::scalar::inverted::{SCORE_COL, SCORE_FIELD};
 use lance_index::vector::{DIST_COL, Query};
-use lance_index::{DatasetIndexExt, scalar::expression::ScalarIndexExpr};
 use lance_index::{metrics::NoOpMetricsCollector, scalar::inverted::FTS_SCHEMA};
 use lance_io::stream::RecordBatchStream;
 use lance_linalg::distance::MetricType;
diff --git a/rust/lance/src/dataset/schema_evolution.rs b/rust/lance/src/dataset/schema_evolution.rs
index b79cb283956..269093fc98b 100644
--- a/rust/lance/src/dataset/schema_evolution.rs
+++ b/rust/lance/src/dataset/schema_evolution.rs
@@ -1783,11 +1783,12 @@ mod test {
     ) -> Result<()> {
         // Create a table with 2 scalar columns, 1 vector column
 
+        use crate::index::DatasetIndexExt;
         use arrow::datatypes::{Int32Type, Int64Type};
         use arrow_array::{Float16Array, Float32Array, Int64Array, ListArray};
         use half::f16;
         use lance_arrow::FixedSizeListArrayExt;
-        use lance_index::{DatasetIndexExt, IndexType, scalar::ScalarIndexParams};
+        use lance_index::{IndexType, scalar::ScalarIndexParams};
         use lance_linalg::distance::MetricType;
         use lance_testing::datagen::generate_random_array;
 
diff --git a/rust/lance/src/dataset/tests/dataset_aggregate.rs b/rust/lance/src/dataset/tests/dataset_aggregate.rs
index cdc954a9997..ef2a90e6315 100644
--- a/rust/lance/src/dataset/tests/dataset_aggregate.rs
+++ b/rust/lance/src/dataset/tests/dataset_aggregate.rs
@@ -38,12 +38,13 @@ use tempfile::tempdir;
 
 use crate::Dataset;
 use crate::dataset::scanner::AggregateExpr;
+use crate::index::DatasetIndexExt;
 use crate::index::vector::VectorIndexParams;
 use crate::utils::test::{DatagenExt, FragmentCount, FragmentRowCount, assert_plan_node_equals};
 use lance_arrow::FixedSizeListArrayExt;
+use lance_index::IndexType;
 use lance_index::scalar::FullTextSearchQuery;
 use lance_index::scalar::inverted::InvertedIndexParams;
-use lance_index::{DatasetIndexExt, IndexType};
 use lance_linalg::distance::MetricType;
 
 /// Helper to create a field reference expression for a column index
diff --git a/rust/lance/src/dataset/tests/dataset_concurrency_store.rs b/rust/lance/src/dataset/tests/dataset_concurrency_store.rs
index 7ce57a4c0f4..a9c2aa44c38 100644
--- a/rust/lance/src/dataset/tests/dataset_concurrency_store.rs
+++ b/rust/lance/src/dataset/tests/dataset_concurrency_store.rs
@@ -8,12 +8,12 @@ use crate::dataset::WriteDestination;
 use crate::{Dataset, Error, Result};
 
 use crate::dataset::write::{WriteMode, WriteParams};
+use crate::index::DatasetIndexExt;
 use arrow_array::RecordBatch;
 use arrow_array::{Int32Array, RecordBatchIterator};
 use arrow_schema::{DataType, Field as ArrowField, Schema as ArrowSchema};
 use futures::TryStreamExt;
 use lance_core::utils::tempfile::TempStrDir;
-use lance_index::DatasetIndexExt;
 use lance_index::{IndexType, scalar::ScalarIndexParams};
 
 #[tokio::test]
diff --git a/rust/lance/src/dataset/tests/dataset_geo.rs b/rust/lance/src/dataset/tests/dataset_geo.rs
index dc1e79dc455..a43718dd7d4 100644
--- a/rust/lance/src/dataset/tests/dataset_geo.rs
+++ b/rust/lance/src/dataset/tests/dataset_geo.rs
@@ -7,6 +7,7 @@ use std::vec;
 use crate::Dataset;
 use crate::dataset::tests::dataset_transactions::execute_sql;
 
+use crate::index::DatasetIndexExt;
 use arrow_array::RecordBatch;
 use arrow_array::RecordBatchIterator;
 use arrow_array::cast::AsArray;
@@ -19,8 +20,8 @@ use geoarrow_array::{
 };
 use geoarrow_schema::{Dimension, LineStringType, PointType, PolygonType};
 use lance_core::utils::tempfile::TempStrDir;
+use lance_index::IndexType;
 use lance_index::scalar::ScalarIndexParams;
-use lance_index::{DatasetIndexExt, IndexType};
 
 #[tokio::test]
 async fn test_geo_types() {
diff --git a/rust/lance/src/dataset/tests/dataset_index.rs b/rust/lance/src/dataset/tests/dataset_index.rs
index c3aac4493d4..a2d4889dd06 100644
--- a/rust/lance/src/dataset/tests/dataset_index.rs
+++ b/rust/lance/src/dataset/tests/dataset_index.rs
@@ -13,6 +13,7 @@ use crate::{Dataset, Error, Result};
 use lance_arrow::FixedSizeListArrayExt;
 
 use crate::dataset::write::{WriteMode, WriteParams};
+use crate::index::DatasetIndexExt;
 use arrow::array::{AsArray, GenericListBuilder, GenericStringBuilder};
 use arrow::datatypes::UInt64Type;
 use arrow_array::RecordBatch;
@@ -31,7 +32,6 @@ use lance_core::utils::tempfile::TempStrDir;
 use lance_datagen::{BatchCount, Dimension, RowCount, array, gen_batch};
 use lance_file::reader::{FileReader, FileReaderOptions};
 use lance_file::version::LanceFileVersion;
-use lance_index::DatasetIndexExt;
 use lance_index::scalar::FullTextSearchQuery;
 use lance_index::scalar::inverted::{
     query::{BooleanQuery, MatchQuery, Occur, Operator, PhraseQuery},
diff --git a/rust/lance/src/dataset/tests/dataset_io.rs b/rust/lance/src/dataset/tests/dataset_io.rs
index e438e0801ea..86ae416b1be 100644
--- a/rust/lance/src/dataset/tests/dataset_io.rs
+++ b/rust/lance/src/dataset/tests/dataset_io.rs
@@ -36,9 +36,10 @@ use lance_file::version::LanceFileVersion;
 use lance_io::assert_io_eq;
 use lance_table::feature_flags;
 
+use crate::index::DatasetIndexExt;
 use futures::TryStreamExt;
+use lance_index::IndexType;
 use lance_index::scalar::ScalarIndexParams;
-use lance_index::{DatasetIndexExt, IndexType};
 use lance_io::object_store::{ObjectStore, ObjectStoreParams};
 use lance_io::utils::tracking_store::IOTracker;
 use lance_table::io::manifest::read_manifest;
diff --git a/rust/lance/src/dataset/tests/dataset_merge_update.rs b/rust/lance/src/dataset/tests/dataset_merge_update.rs
index ef64773cc3c..1bdee1227b1 100644
--- a/rust/lance/src/dataset/tests/dataset_merge_update.rs
+++ b/rust/lance/src/dataset/tests/dataset_merge_update.rs
@@ -9,11 +9,12 @@ use crate::dataset::WriteDestination;
 use crate::dataset::optimize::{CompactionOptions, compact_files};
 use crate::dataset::transaction::{DataReplacementGroup, Operation};
 use crate::dataset::{AutoCleanupParams, MergeInsertBuilder, ProjectionRequest};
+use crate::index::DatasetIndexExt;
 use crate::{Dataset, Error};
 use lance_core::ROW_ADDR;
+use lance_index::IndexType;
 use lance_index::optimize::OptimizeOptions;
 use lance_index::scalar::ScalarIndexParams;
-use lance_index::{DatasetIndexExt, IndexType};
 use mock_instant::thread_local::MockClock;
 
 use crate::dataset::write::{InsertBuilder, WriteMode, WriteParams};
diff --git a/rust/lance/src/dataset/tests/dataset_migrations.rs b/rust/lance/src/dataset/tests/dataset_migrations.rs
index ecbffa717ba..d71a65bfa69 100644
--- a/rust/lance/src/dataset/tests/dataset_migrations.rs
+++ b/rust/lance/src/dataset/tests/dataset_migrations.rs
@@ -11,12 +11,12 @@ use crate::{Dataset, Result};
 use lance_table::format::IndexMetadata;
 
 use crate::dataset::write::{WriteMode, WriteParams};
+use crate::index::DatasetIndexExt;
 use arrow::compute::concat_batches;
 use arrow_array::RecordBatch;
 use arrow_array::{Float32Array, Int64Array, RecordBatchIterator};
 use arrow_schema::Schema as ArrowSchema;
 use lance_file::version::LanceFileVersion;
-use lance_index::DatasetIndexExt;
 
 use futures::{StreamExt, TryStreamExt};
 use rstest::rstest;
diff --git a/rust/lance/src/dataset/tests/dataset_scanner.rs b/rust/lance/src/dataset/tests/dataset_scanner.rs
index e5e72b48f84..d5dac4c8562 100644
--- a/rust/lance/src/dataset/tests/dataset_scanner.rs
+++ b/rust/lance/src/dataset/tests/dataset_scanner.rs
@@ -9,6 +9,7 @@ use crate::index::vector::VectorIndexParams;
 use lance_arrow::FixedSizeListArrayExt;
 use lance_arrow::json::{JsonArray, is_arrow_json_field, json_field};
 
+use crate::index::DatasetIndexExt;
 use arrow::compute::concat_batches;
 use arrow_array::UInt64Array;
 use arrow_array::{Array, FixedSizeListArray};
@@ -24,7 +25,7 @@ use lance_index::scalar::FullTextSearchQuery;
 use lance_index::scalar::inverted::{
     SCORE_FIELD, query::PhraseQuery, tokenizer::InvertedIndexParams,
 };
-use lance_index::{DatasetIndexExt, IndexType, vector::DIST_COL};
+use lance_index::{IndexType, vector::DIST_COL};
 use lance_io::scheduler::{ScanScheduler, SchedulerConfig};
 use lance_io::utils::CachedFileSize;
 use lance_linalg::distance::MetricType;
diff --git a/rust/lance/src/dataset/tests/dataset_transactions.rs b/rust/lance/src/dataset/tests/dataset_transactions.rs
index 1f522800a44..2b49d0963e1 100644
--- a/rust/lance/src/dataset/tests/dataset_transactions.rs
+++ b/rust/lance/src/dataset/tests/dataset_transactions.rs
@@ -14,13 +14,13 @@ use crate::{Dataset, Result};
 use lance_table::io::commit::ManifestNamingScheme;
 
 use crate::dataset::write::{CommitBuilder, InsertBuilder, WriteMode, WriteParams};
+use crate::index::DatasetIndexExt;
 use arrow_array::Array;
 use arrow_array::RecordBatch;
 use arrow_array::{Int32Array, RecordBatchIterator, StringArray, types::Int32Type};
 use arrow_schema::{DataType, Field as ArrowField, Schema as ArrowSchema};
 use lance_core::utils::tempfile::{TempDir, TempStrDir};
 use lance_datagen::{BatchCount, RowCount, array};
-use lance_index::DatasetIndexExt;
 
 use crate::datafusion::LanceTableProvider;
 use datafusion::prelude::SessionContext;
diff --git a/rust/lance/src/dataset/udtf.rs b/rust/lance/src/dataset/udtf.rs
index d25aec45009..b39586c2777 100644
--- a/rust/lance/src/dataset/udtf.rs
+++ b/rust/lance/src/dataset/udtf.rs
@@ -259,13 +259,14 @@ impl FtsQueryUDTFBuilder {
 pub mod tests {
     use crate::Dataset;
     use crate::dataset::udtf::FtsQueryUDTFBuilder;
+    use crate::index::DatasetIndexExt;
     use arrow_array::{
         Array, Int32Array, RecordBatch, RecordBatchIterator, StringArray, UInt64Array,
     };
     use arrow_schema::{DataType, Field};
     use datafusion::prelude::SessionContext;
+    use lance_index::IndexType;
     use lance_index::scalar::InvertedIndexParams;
-    use lance_index::{DatasetIndexExt, IndexType};
     use std::sync::Arc;
 
     #[tokio::test]
diff --git a/rust/lance/src/dataset/write/delete.rs b/rust/lance/src/dataset/write/delete.rs
index 18916b087f5..2f7314925e6 100644
--- a/rust/lance/src/dataset/write/delete.rs
+++ b/rust/lance/src/dataset/write/delete.rs
@@ -288,6 +288,7 @@ mod tests {
     use super::*;
     use crate::dataset::{InsertBuilder, UpdateBuilder};
     use crate::dataset::{WriteMode, WriteParams};
+    use crate::index::DatasetIndexExt;
     use crate::utils::test::TestDatasetGenerator;
     use arrow::array::AsArray;
     use arrow::datatypes::UInt32Type;
@@ -296,7 +297,7 @@ mod tests {
     use futures::TryStreamExt;
     use lance_core::utils::tempfile::TempStrDir;
     use lance_file::version::LanceFileVersion;
-    use lance_index::{DatasetIndexExt, IndexType, scalar::ScalarIndexParams};
+    use lance_index::{IndexType, scalar::ScalarIndexParams};
     use rstest::rstest;
     use std::collections::HashSet;
     use std::ops::Range;
diff --git a/rust/lance/src/dataset/write/merge_insert.rs b/rust/lance/src/dataset/write/merge_insert.rs
index cfc1e8f0dca..2c56d8349fc 100644
--- a/rust/lance/src/dataset/write/merge_insert.rs
+++ b/rust/lance/src/dataset/write/merge_insert.rs
@@ -29,6 +29,7 @@ use super::{CommitBuilder, WriteParams, write_fragments_internal};
 use crate::dataset::rowids::get_row_id_index;
 use crate::dataset::transaction::UpdateMode::{RewriteColumns, RewriteRows};
 use crate::dataset::utils::CapturedRowIds;
+use crate::index::DatasetIndexExt;
 use crate::{
     Dataset,
     datafusion::dataframe::SessionContextExt,
@@ -94,8 +95,8 @@ use lance_datafusion::{
     utils::StreamingWriteSource,
 };
 use lance_file::version::LanceFileVersion;
+use lance_index::IndexCriteria;
 use lance_index::mem_wal::MergedGeneration;
-use lance_index::{DatasetIndexExt, IndexCriteria};
 use lance_table::format::{Fragment, IndexMetadata, RowIdMeta};
 use log::info;
 use roaring::RoaringTreemap;
diff --git a/rust/lance/src/dataset/write/update.rs b/rust/lance/src/dataset/write/update.rs
index 855d83018f4..ec34000642d 100644
--- a/rust/lance/src/dataset/write/update.rs
+++ b/rust/lance/src/dataset/write/update.rs
@@ -483,6 +483,7 @@ mod tests {
     use super::*;
 
     use crate::dataset::{WriteDestination, WriteMode};
+    use crate::index::DatasetIndexExt;
     use crate::index::vector::VectorIndexParams;
     use crate::utils::test::{DatagenExt, FragmentCount, FragmentRowCount};
     use arrow::{array::AsArray, datatypes::UInt32Type};
@@ -495,7 +496,6 @@ mod tests {
     use lance_core::utils::tempfile::TempStrDir;
     use lance_datagen::{Dimension, RowCount};
     use lance_file::version::LanceFileVersion;
-    use lance_index::DatasetIndexExt;
     use lance_index::IndexType;
     use lance_index::scalar::ScalarIndexParams;
     use lance_io::object_store::ObjectStoreParams;
diff --git a/rust/lance/src/index.rs b/rust/lance/src/index.rs
index 28da8f61e1b..843fc8c7740 100644
--- a/rust/lance/src/index.rs
+++ b/rust/lance/src/index.rs
@@ -24,6 +24,7 @@ use lance_core::utils::tracing::{
 };
 use lance_file::previous::reader::FileReader as PreviousFileReader;
 use lance_file::reader::FileReaderOptions;
+use lance_index::INDEX_METADATA_SCHEMA_KEY;
 pub use lance_index::IndexParams;
 use lance_index::frag_reuse::{FRAG_REUSE_INDEX_NAME, FragReuseIndex};
 use lance_index::mem_wal::{MEM_WAL_INDEX_NAME, MemWalIndex};
@@ -42,7 +43,6 @@ use lance_index::vector::flat::index::{FlatBinQuantizer, FlatIndex, FlatQuantize
 use lance_index::vector::hnsw::HNSW;
 use lance_index::vector::pq::ProductQuantizer;
 use lance_index::vector::sq::ScalarQuantizer;
-use lance_index::{DatasetIndexExt, INDEX_METADATA_SCHEMA_KEY, IndexDescription, IndexSegment};
 use lance_index::{INDEX_FILE_NAME, Index, IndexType, pb, vector::VectorIndex};
 use lance_index::{
     IndexCriteria, is_system_index,
@@ -65,6 +65,7 @@ use uuid::Uuid;
 use vector::ivf::v2::IVFIndex;
 use vector::utils::get_vector_type;
 
+mod api;
 pub(crate) mod append;
 mod create;
 pub mod frag_reuse;
@@ -79,6 +80,7 @@ use crate::dataset::index::LanceIndexStoreExt;
 use crate::dataset::optimize::RemappedIndex;
 use crate::dataset::optimize::remapping::RemapResult;
 use crate::dataset::transaction::{Operation, Transaction, TransactionBuilder};
+pub use crate::index::api::{DatasetIndexExt, IndexSegment, IndexSegmentPlan};
 use crate::index::frag_reuse::{load_frag_reuse_index_details, open_frag_reuse_index};
 use crate::index::mem_wal::open_mem_wal_index;
 pub use crate::index::prefilter::{FilterLoader, PreFilter};
@@ -86,6 +88,7 @@ use crate::index::scalar::{IndexDetails, fetch_index_details, load_training_data
 use crate::session::index_caches::{FragReuseIndexKey, IndexMetadataKey};
 use crate::{Error, Result, dataset::Dataset};
 pub use create::CreateIndexBuilder;
+pub use lance_index::IndexDescription;
 
 fn validate_index_segments(index_name: &str, segments: &[IndexSegment]) -> Result<()> {
     if segments.is_empty() {
@@ -645,7 +648,8 @@ impl DatasetIndexExt for Dataset {
     /// Create a scalar BTREE index:
     /// ```
     /// # use lance::{Dataset, Result};
-    /// # use lance_index::{DatasetIndexExt, IndexType, scalar::ScalarIndexParams};
+    /// # use lance::index::DatasetIndexExt;
+    /// # use lance_index::{IndexType, scalar::ScalarIndexParams};
     /// # async fn example(dataset: &mut Dataset) -> Result<()> {
     /// let params = ScalarIndexParams::default();
     /// dataset
@@ -659,7 +663,8 @@ impl DatasetIndexExt for Dataset {
     /// Create an empty index that will be populated later:
     /// ```
     /// # use lance::{Dataset, Result};
-    /// # use lance_index::{DatasetIndexExt, IndexType, scalar::ScalarIndexParams};
+    /// # use lance::index::DatasetIndexExt;
+    /// # use lance_index::{IndexType, scalar::ScalarIndexParams};
     /// # async fn example(dataset: &mut Dataset) -> Result<()> {
     /// let params = ScalarIndexParams::default();
     /// dataset
diff --git a/rust/lance/src/index/api.rs b/rust/lance/src/index/api.rs
new file mode 100644
index 00000000000..f8e7ee7d012
--- /dev/null
+++ b/rust/lance/src/index/api.rs
@@ -0,0 +1,294 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The Lance Authors
+
+use std::sync::Arc;
+
+use async_trait::async_trait;
+use datafusion::execution::SendableRecordBatchStream;
+use lance_index::{IndexParams, IndexType, optimize::OptimizeOptions};
+use lance_table::format::IndexMetadata;
+use roaring::RoaringBitmap;
+use uuid::Uuid;
+
+use crate::{Error, Result};
+
+/// A single physical segment of a logical index.
+///
+/// Each segment is stored independently and will become one manifest entry when committed.
+/// The logical index identity (name / target column / dataset version) is provided separately
+/// by the commit API.
+#[derive(Debug, Clone, PartialEq)]
+pub struct IndexSegment {
+    /// Unique ID of the physical segment.
+    uuid: Uuid,
+    /// The fragments covered by this segment.
+    fragment_bitmap: RoaringBitmap,
+    /// Metadata specific to the index type.
+    index_details: Arc<prost_types::Any>,
+    /// The on-disk index version for this segment.
+    index_version: i32,
+}
+
+impl IndexSegment {
+    /// Create a fully described segment with the given UUID, fragment coverage, and index
+    /// metadata.
+    pub fn new<I>(
+        uuid: Uuid,
+        fragment_bitmap: I,
+        index_details: Arc<prost_types::Any>,
+        index_version: i32,
+    ) -> Self
+    where
+        I: IntoIterator<Item = u32>,
+    {
+        Self {
+            uuid,
+            fragment_bitmap: fragment_bitmap.into_iter().collect(),
+            index_details,
+            index_version,
+        }
+    }
+
+    /// Return the UUID of this segment.
+    pub fn uuid(&self) -> Uuid {
+        self.uuid
+    }
+
+    /// Return the fragment coverage of this segment.
+    pub fn fragment_bitmap(&self) -> &RoaringBitmap {
+        &self.fragment_bitmap
+    }
+
+    /// Return the serialized index details for this segment.
+    pub fn index_details(&self) -> &Arc<prost_types::Any> {
+        &self.index_details
+    }
+
+    /// Return the on-disk index version for this segment.
+    pub fn index_version(&self) -> i32 {
+        self.index_version
+    }
+
+    /// Consume the segment and return its component parts.
+    pub fn into_parts(self) -> (Uuid, RoaringBitmap, Arc<prost_types::Any>, i32) {
+        (
+            self.uuid,
+            self.fragment_bitmap,
+            self.index_details,
+            self.index_version,
+        )
+    }
+}
+
+/// A plan for building one physical segment from one or more existing
+/// vector index segments.
+#[derive(Debug, Clone, PartialEq)]
+pub struct IndexSegmentPlan {
+    segment: IndexSegment,
+    segments: Vec<IndexMetadata>,
+    estimated_bytes: u64,
+    requested_index_type: Option<IndexType>,
+}
+
+impl IndexSegmentPlan {
+    /// Create a plan for one built segment.
+    pub fn new(
+        segment: IndexSegment,
+        segments: Vec<IndexMetadata>,
+        estimated_bytes: u64,
+        requested_index_type: Option<IndexType>,
+    ) -> Self {
+        Self {
+            segment,
+            segments,
+            estimated_bytes,
+            requested_index_type,
+        }
+    }
+
+    /// Return the segment metadata that should be committed after this plan is built.
+    pub fn segment(&self) -> &IndexSegment {
+        &self.segment
+    }
+
+    /// Return the input segment metadata that should be combined into the segment.
+    pub fn segments(&self) -> &[IndexMetadata] {
+        &self.segments
+    }
+
+    /// Return the estimated number of bytes covered by this plan.
+    pub fn estimated_bytes(&self) -> u64 {
+        self.estimated_bytes
+    }
+
+    /// Return the requested logical index type, if one was supplied to the planner.
+    pub fn requested_index_type(&self) -> Option<IndexType> {
+        self.requested_index_type
+    }
+}
+
+/// Extends [`crate::Dataset`] with secondary index APIs.
+#[async_trait]
+pub trait DatasetIndexExt {
+    type IndexBuilder<'a>
+    where
+        Self: 'a;
+    type IndexSegmentBuilder<'a>
+    where
+        Self: 'a;
+
+    /// Create a builder for creating an index on columns.
+    ///
+    /// This returns a builder that can be configured with additional options
+    /// like `name()`, `replace()`, and `train()` before awaiting to execute.
+    fn create_index_builder<'a>(
+        &'a mut self,
+        columns: &'a [&'a str],
+        index_type: IndexType,
+        params: &'a dyn IndexParams,
+    ) -> Self::IndexBuilder<'a>;
+
+    /// Create a builder for building physical index segments from uncommitted
+    /// vector index outputs.
+    ///
+    /// The caller supplies the uncommitted index metadata returned by
+    /// `execute_uncommitted()` so the builder can plan segment grouping without
+    /// rediscovering fragment coverage.
+    ///
+    /// This is the canonical entry point for distributed vector segment build.
+    /// After building the physical segments, publish them as a
+    /// logical index with [`Self::commit_existing_index_segments`].
+    fn create_index_segment_builder<'a>(&'a self) -> Self::IndexSegmentBuilder<'a>;
+
+    /// Create indices on columns.
+    ///
+    /// Upon finish, a new dataset version is generated.
+    async fn create_index(
+        &mut self,
+        columns: &[&str],
+        index_type: IndexType,
+        name: Option<String>,
+        params: &dyn IndexParams,
+        replace: bool,
+    ) -> Result<IndexMetadata>;
+
+    /// Drop indices by name.
+    ///
+    /// Upon finish, a new dataset version is generated.
+    async fn drop_index(&mut self, name: &str) -> Result<()>;
+
+    /// Prewarm an index by name.
+    ///
+    /// This will load the index into memory and cache it.
+    async fn prewarm_index(&self, name: &str) -> Result<()>;
+
+    /// Read all indices of this Dataset version.
+    ///
+    /// The indices are lazy loaded and cached in memory within the `Dataset` instance.
+    /// The cache is invalidated when the dataset version (Manifest) is changed.
+    async fn load_indices(&self) -> Result<Arc<Vec<IndexMetadata>>>;
+
+    /// Loads all the indices of a given UUID.
+    ///
+    /// Note that it is possible to have multiple indices with the same UUID,
+    /// as they are the deltas of the same index.
+    async fn load_index(&self, uuid: &str) -> Result<Option<IndexMetadata>> {
+        self.load_indices().await.map(|indices| {
+            indices
+                .iter()
+                .find(|idx| idx.uuid.to_string() == uuid)
+                .cloned()
+        })
+    }
+
+    /// Loads a specific index with the given index name.
+    ///
+    /// Returns `Ok(vec![])` if the index does not exist.
+    async fn load_indices_by_name(&self, name: &str) -> Result<Vec<IndexMetadata>> {
+        self.load_indices().await.map(|indices| {
+            indices
+                .iter()
+                .filter(|idx| idx.name == name)
+                .cloned()
+                .collect()
+        })
+    }
+
+    /// Loads a specific index with the given index name.
+    /// This function only works for indices that are unique.
+    /// If there are multiple indices sharing the same name, please use [`Self::load_indices_by_name`].
+    async fn load_index_by_name(&self, name: &str) -> Result<Option<IndexMetadata>> {
+        let indices = self.load_indices_by_name(name).await?;
+        if indices.is_empty() {
+            Ok(None)
+        } else if indices.len() == 1 {
+            Ok(Some(indices[0].clone()))
+        } else {
+            Err(Error::index(format!(
+                "Found multiple indices of the same name: {:?}, please use load_indices_by_name",
+                indices.iter().map(|idx| &idx.name).collect::<Vec<_>>()
+            )))
+        }
+    }
+
+    /// Describes indexes in a dataset.
+    ///
+    /// This method should only access the index metadata and should not load the index into memory.
+    async fn describe_indices<'a, 'b>(
+        &'a self,
+        criteria: Option<lance_index::IndexCriteria<'b>>,
+    ) -> Result<Vec<Arc<dyn lance_index::IndexDescription>>>;
+
+    /// Loads a specific scalar index using the provided criteria.
+    async fn load_scalar_index<'a, 'b>(
+        &'a self,
+        criteria: lance_index::IndexCriteria<'b>,
+    ) -> Result<Option<IndexMetadata>>;
+
+    /// Optimize indices.
+    async fn optimize_indices(&mut self, options: &OptimizeOptions) -> Result<()>;
+
+    /// Find an index with the given name and return its serialized statistics.
+    async fn index_statistics(&self, index_name: &str) -> Result<String>;
+
+    /// Commit one or more existing physical index segments as a logical index.
+    async fn commit_existing_index_segments(
+        &mut self,
+        index_name: &str,
+        column: &str,
+        segments: Vec<IndexSegment>,
+    ) -> Result<()>;
+
+    async fn read_index_partition(
+        &self,
+        index_name: &str,
+        partition_id: usize,
+        with_vector: bool,
+    ) -> Result<SendableRecordBatchStream>;
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use super::{IndexSegment, IndexSegmentPlan};
+    use lance_index::IndexType;
+    use uuid::Uuid;
+
+    #[test]
+    fn test_index_segment_plan_accessors() {
+        let uuid = Uuid::new_v4();
+        let segment = IndexSegment::new(uuid, [1_u32, 3], Arc::new(prost_types::Any::default()), 7);
+        let plan = IndexSegmentPlan::new(segment.clone(), vec![], 128, Some(IndexType::BTree));
+
+        assert_eq!(segment.uuid(), uuid);
+        assert_eq!(
+            segment.fragment_bitmap().iter().collect::<Vec<_>>(),
+            vec![1, 3]
+        );
+        assert_eq!(segment.index_version(), 7);
+        assert_eq!(plan.segment().uuid(), uuid);
+        assert_eq!(plan.estimated_bytes(), 128);
+        assert_eq!(plan.requested_index_type(), Some(IndexType::BTree));
+    }
+}
diff --git a/rust/lance/src/index/append.rs b/rust/lance/src/index/append.rs
index b9944b9e656..d80b9bf710e 100644
--- a/rust/lance/src/index/append.rs
+++ b/rust/lance/src/index/append.rs
@@ -310,6 +310,7 @@ pub async fn merge_indices_with_unindexed_frags<'a>(
 mod tests {
     use super::*;
 
+    use crate::index::DatasetIndexExt;
     use arrow::datatypes::{Float32Type, UInt32Type};
     use arrow_array::cast::AsArray;
     use arrow_array::{
@@ -324,7 +325,7 @@ mod tests {
     use lance_index::vector::hnsw::builder::HnswBuildParams;
     use lance_index::vector::sq::builder::SQBuildParams;
     use lance_index::{
-        DatasetIndexExt, IndexType,
+        IndexType,
         scalar::ScalarIndexParams,
         vector::{ivf::IvfBuildParams, pq::PQBuildParams},
     };
diff --git a/rust/lance/src/index/create.rs b/rust/lance/src/index/create.rs
index a394f52258e..0bf9fdd283c 100644
--- a/rust/lance/src/index/create.rs
+++ b/rust/lance/src/index/create.rs
@@ -20,7 +20,7 @@ use crate::{
 use futures::future::{BoxFuture, try_join_all};
 use lance_core::datatypes::format_field_path;
 use lance_index::progress::{IndexBuildProgress, NoopIndexBuildProgress};
-use lance_index::{IndexParams, IndexSegment, IndexSegmentPlan, IndexType, scalar::CreatedIndex};
+use lance_index::{IndexParams, IndexType, scalar::CreatedIndex};
 use lance_index::{
     metrics::NoOpMetricsCollector,
     scalar::{LANCE_SCALAR_INDEX, ScalarIndexParams, inverted::tokenizer::InvertedIndexParams},
@@ -32,6 +32,8 @@ use uuid::Uuid;
 
 use arrow_array::RecordBatchReader;
 
+use super::{IndexSegment, IndexSegmentPlan};
+
 /// Generate default index name from field path.
 ///
 /// Joins field names with `.` to create the base index name.
diff --git a/rust/lance/src/index/frag_reuse.rs b/rust/lance/src/index/frag_reuse.rs
index e9e66f972b6..e5f63514d86 100644
--- a/rust/lance/src/index/frag_reuse.rs
+++ b/rust/lance/src/index/frag_reuse.rs
@@ -3,8 +3,8 @@
 
 use crate::Dataset;
 use crate::dataset::optimize::remapping::transpose_row_ids_from_digest;
+use crate::index::DatasetIndexExt;
 use lance_core::Error;
-use lance_index::DatasetIndexExt;
 use lance_index::frag_reuse::{
     FRAG_REUSE_DETAILS_FILE_NAME, FRAG_REUSE_INDEX_NAME, FragReuseGroup, FragReuseIndex,
     FragReuseIndexDetails, FragReuseVersion,
diff --git a/rust/lance/src/index/mem_wal.rs b/rust/lance/src/index/mem_wal.rs
index c979ab42fe2..66d3a9aaca2 100644
--- a/rust/lance/src/index/mem_wal.rs
+++ b/rust/lance/src/index/mem_wal.rs
@@ -122,9 +122,9 @@ mod tests {
 
     use std::sync::Arc;
 
+    use crate::index::DatasetIndexExt;
     use arrow_array::{Int32Array, RecordBatch};
     use arrow_schema::{DataType, Field, Schema};
-    use lance_index::DatasetIndexExt;
 
     use crate::dataset::transaction::{Operation, Transaction};
     use crate::dataset::{CommitBuilder, InsertBuilder, WriteParams};
diff --git a/rust/lance/src/index/scalar.rs b/rust/lance/src/index/scalar.rs
index f2c9cfd8969..44739454bec 100644
--- a/rust/lance/src/index/scalar.rs
+++ b/rust/lance/src/index/scalar.rs
@@ -6,6 +6,7 @@
 
 use std::sync::{Arc, LazyLock};
 
+use crate::index::DatasetIndexExt;
 use crate::index::DatasetIndexInternalExt;
 use crate::session::index_caches::ProstAny;
 use crate::{
@@ -39,7 +40,7 @@ use lance_index::scalar::{
     ScalarIndex, ScalarIndexParams, bitmap::BITMAP_LOOKUP_NAME, inverted::INVERT_LIST_FILE,
     lance_format::LanceIndexStore,
 };
-use lance_index::{DatasetIndexExt, IndexCriteria, IndexType};
+use lance_index::{IndexCriteria, IndexType};
 use lance_table::format::{Fragment, IndexMetadata};
 use log::info;
 use tracing::instrument;
@@ -819,9 +820,9 @@ mod tests {
     #[tokio::test]
     async fn test_initialize_scalar_index_btree() {
         use crate::dataset::Dataset;
+        use crate::index::DatasetIndexExt;
         use arrow_array::types::Float32Type;
         use lance_datagen::{BatchCount, RowCount, array};
-        use lance_index::DatasetIndexExt;
         use lance_index::metrics::NoOpMetricsCollector;
         use lance_index::scalar::ScalarIndexParams;
 
@@ -925,9 +926,9 @@ mod tests {
     #[tokio::test]
     async fn test_optimize_scalar_index_btree() {
         use crate::dataset::Dataset;
+        use crate::index::DatasetIndexExt;
         use arrow_array::types::Float32Type;
         use lance_datagen::{BatchCount, RowCount, array};
-        use lance_index::DatasetIndexExt;
         use lance_index::metrics::NoOpMetricsCollector;
         use lance_index::scalar::ScalarIndexParams;
 
@@ -1043,9 +1044,9 @@ mod tests {
     #[tokio::test]
     async fn test_initialize_scalar_index_bitmap() {
         use crate::dataset::Dataset;
+        use crate::index::DatasetIndexExt;
         use arrow_array::types::Float32Type;
         use lance_datagen::{BatchCount, RowCount, array};
-        use lance_index::DatasetIndexExt;
         use lance_index::scalar::ScalarIndexParams;
 
         let test_dir = TempStrDir::default();
@@ -1123,8 +1124,8 @@ mod tests {
     #[tokio::test]
     async fn test_initialize_scalar_index_inverted() {
         use crate::dataset::Dataset;
+        use crate::index::DatasetIndexExt;
         use lance_datagen::{BatchCount, ByteCount, RowCount, array};
-        use lance_index::DatasetIndexExt;
         use lance_index::metrics::NoOpMetricsCollector;
         use lance_index::scalar::inverted::tokenizer::InvertedIndexParams;
 
@@ -1262,9 +1263,9 @@ mod tests {
     #[tokio::test]
     async fn test_initialize_scalar_index_zonemap() {
         use crate::dataset::Dataset;
+        use crate::index::DatasetIndexExt;
         use arrow_array::types::Float32Type;
         use lance_datagen::{BatchCount, RowCount, array};
-        use lance_index::DatasetIndexExt;
         use lance_index::metrics::NoOpMetricsCollector;
         use lance_index::scalar::ScalarIndexParams;
         use lance_index::scalar::zonemap::ZoneMapIndexBuilderParams;
diff --git a/rust/lance/src/index/vector.rs b/rust/lance/src/index/vector.rs
index a19f472a9bf..4b5b113cae9 100644
--- a/rust/lance/src/index/vector.rs
+++ b/rust/lance/src/index/vector.rs
@@ -47,9 +47,7 @@ use lance_index::vector::{
     pq::PQBuildParams,
     sq::{ScalarQuantizer, builder::SQBuildParams},
 };
-use lance_index::{
-    DatasetIndexExt, INDEX_AUXILIARY_FILE_NAME, INDEX_METADATA_SCHEMA_KEY, IndexType,
-};
+use lance_index::{INDEX_AUXILIARY_FILE_NAME, INDEX_METADATA_SCHEMA_KEY, IndexType};
 use lance_io::traits::Reader;
 use lance_linalg::distance::*;
 use lance_table::format::{IndexMetadata, list_index_files_with_sizes};
@@ -58,7 +56,7 @@ use tracing::instrument;
 use utils::get_vector_type;
 use uuid::Uuid;
 
-use super::{DatasetIndexInternalExt, IndexParams, pb, vector_index_details};
+use super::{DatasetIndexExt, DatasetIndexInternalExt, IndexParams, pb, vector_index_details};
 use crate::dataset::index::dataset_format_version;
 use crate::dataset::transaction::{Operation, Transaction};
 use crate::{Error, Result, dataset::Dataset, index::pb::vector_index_stage::Stage};
@@ -1688,6 +1686,7 @@ fn derive_hnsw_params(source_index: &dyn VectorIndex) -> HnswBuildParams {
 mod tests {
     use super::*;
     use crate::dataset::Dataset;
+    use crate::index::DatasetIndexExt;
     use arrow_array::Array;
     use arrow_array::RecordBatch;
     use arrow_array::types::{Float32Type, Int32Type};
@@ -1695,7 +1694,6 @@ mod tests {
     use lance_core::utils::tempfile::TempStrDir;
     use lance_datagen::{BatchCount, RowCount, array};
     use lance_file::writer::FileWriterOptions;
-    use lance_index::DatasetIndexExt;
     use lance_index::metrics::NoOpMetricsCollector;
     use lance_linalg::distance::MetricType;
 
diff --git a/rust/lance/src/index/vector/ivf.rs b/rust/lance/src/index/vector/ivf.rs
index 8e47b77cdfd..90864154094 100644
--- a/rust/lance/src/index/vector/ivf.rs
+++ b/rust/lance/src/index/vector/ivf.rs
@@ -64,8 +64,7 @@ use lance_index::vector::quantizer::QuantizationType;
 use lance_index::vector::v3::shuffler::create_ivf_shuffler;
 use lance_index::vector::v3::subindex::{IvfSubIndex, SubIndexType};
 use lance_index::{
-    INDEX_AUXILIARY_FILE_NAME, INDEX_METADATA_SCHEMA_KEY, Index, IndexMetadata, IndexSegment,
-    IndexSegmentPlan, IndexType,
+    INDEX_AUXILIARY_FILE_NAME, INDEX_METADATA_SCHEMA_KEY, Index, IndexMetadata, IndexType,
     optimize::OptimizeOptions,
     vector::{
         Query, VectorIndex,
@@ -103,6 +102,8 @@ use tokio::sync::mpsc;
 use tracing::instrument;
 use uuid::Uuid;
 
+use crate::index::{IndexSegment, IndexSegmentPlan};
+
 pub mod builder;
 pub mod io;
 pub mod v2;
diff --git a/rust/lance/src/index/vector/ivf/v2.rs b/rust/lance/src/index/vector/ivf/v2.rs
index 92b0549f43b..70ee762f59d 100644
--- a/rust/lance/src/index/vector/ivf/v2.rs
+++ b/rust/lance/src/index/vector/ivf/v2.rs
@@ -626,7 +626,9 @@ mod tests {
     use lance_index::vector::storage::VectorStore;
 
     use crate::dataset::{InsertBuilder, UpdateBuilder, WriteMode, WriteParams};
+    use crate::index::DatasetIndexExt;
     use crate::index::DatasetIndexInternalExt;
+    use crate::index::IndexSegment;
     use crate::index::vector::ivf::v2::IvfPq;
     use crate::index::vector::ivf::{build_segment, plan_segments};
     use crate::utils::test::copy_test_data_to_tmp;
@@ -644,6 +646,7 @@ mod tests {
     use lance_encoding::decoder::DecoderPlugins;
     use lance_file::reader::{FileReader, FileReaderOptions};
     use lance_file::writer::FileWriter;
+    use lance_index::IndexType;
     use lance_index::vector::DIST_COL;
     use lance_index::vector::ivf::IvfBuildParams;
     use lance_index::vector::kmeans::{KMeansParams, train_kmeans};
@@ -653,7 +656,6 @@ mod tests {
     use lance_index::vector::{
         pq::storage::ProductQuantizationMetadata, storage::STORAGE_METADATA_KEY,
     };
-    use lance_index::{DatasetIndexExt, IndexSegment, IndexType};
     use lance_index::{INDEX_AUXILIARY_FILE_NAME, metrics::NoOpMetricsCollector};
     use lance_index::{optimize::OptimizeOptions, scalar::IndexReader};
     use lance_index::{scalar::IndexWriter, vector::hnsw::builder::HnswBuildParams};
diff --git a/rust/lance/src/io/commit.rs b/rust/lance/src/io/commit.rs
index 7a7b9a5a77a..2e61054d28f 100644
--- a/rust/lance/src/io/commit.rs
+++ b/rust/lance/src/io/commit.rs
@@ -48,6 +48,7 @@ use crate::dataset::{
     ManifestWriteConfig, NewTransactionResult, TRANSACTIONS_DIR, load_new_transactions,
     write_manifest_file,
 };
+use crate::index::DatasetIndexExt;
 use crate::index::DatasetIndexInternalExt;
 use crate::io::deletion::read_dataset_deletion_file;
 use crate::session::Session;
@@ -56,7 +57,7 @@ use crate::session::index_caches::IndexMetadataKey;
 use futures::future::Either;
 use futures::{StreamExt, TryFutureExt, TryStreamExt};
 use lance_core::{Error, Result};
-use lance_index::{DatasetIndexExt, is_system_index};
+use lance_index::is_system_index;
 use lance_io::object_store::ObjectStoreRegistry;
 use log;
 use object_store::path::Path;
diff --git a/rust/lance/src/io/commit/conflict_resolver.rs b/rust/lance/src/io/commit/conflict_resolver.rs
index 3765b87eec7..8949cb383cc 100644
--- a/rust/lance/src/io/commit/conflict_resolver.rs
+++ b/rust/lance/src/io/commit/conflict_resolver.rs
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The Lance Authors
 
+use crate::index::DatasetIndexExt;
 use crate::index::frag_reuse::{build_frag_reuse_index_metadata, load_frag_reuse_index_details};
 use crate::index::mem_wal::{load_mem_wal_index_details, new_mem_wal_index_meta};
 use crate::io::deletion::read_dataset_deletion_file;
@@ -14,7 +15,6 @@ use lance_core::{
     Error, Result,
     utils::{deletion::DeletionVector, mask::RowAddrTreeMap},
 };
-use lance_index::DatasetIndexExt;
 use lance_index::frag_reuse::FRAG_REUSE_INDEX_NAME;
 use lance_index::mem_wal::{MEM_WAL_INDEX_NAME, MergedGeneration};
 use lance_table::format::IndexMetadata;
diff --git a/rust/lance/src/io/exec/filtered_read.rs b/rust/lance/src/io/exec/filtered_read.rs
index 35fd486a283..e3837f9ce4f 100644
--- a/rust/lance/src/io/exec/filtered_read.rs
+++ b/rust/lance/src/io/exec/filtered_read.rs
@@ -2052,6 +2052,7 @@ impl ExecutionPlan for FilteredReadExec {
 mod tests {
     use std::collections::HashSet;
 
+    use crate::index::DatasetIndexExt;
     use arrow::{
         compute::concat_batches,
         datatypes::{Float32Type, UInt32Type, UInt64Type},
@@ -2064,7 +2065,7 @@ mod tests {
     use lance_core::utils::tempfile::TempStrDir;
     use lance_datagen::{BatchCount, Dimension, RowCount, array, gen_batch};
     use lance_index::{
-        DatasetIndexExt, IndexType,
+        IndexType,
         optimize::OptimizeOptions,
         scalar::{ScalarIndexParams, expression::PlannerIndexExt},
     };
diff --git a/rust/lance/src/io/exec/fts.rs b/rust/lance/src/io/exec/fts.rs
index 6e129841c76..0858c294753 100644
--- a/rust/lance/src/io/exec/fts.rs
+++ b/rust/lance/src/io/exec/fts.rs
@@ -25,7 +25,9 @@ use lance_datafusion::utils::{ExecutionPlanMetricsSetExt, MetricsExt, PARTITIONS
 
 use super::PreFilterSource;
 use super::utils::{IndexMetrics, InstrumentedRecordBatchStreamAdapter, build_prefilter};
+use crate::index::DatasetIndexExt;
 use crate::{Dataset, index::DatasetIndexInternalExt};
+use lance_index::IndexCriteria;
 use lance_index::metrics::MetricsCollector;
 use lance_index::scalar::inverted::builder::document_input;
 use lance_index::scalar::inverted::lance_tokenizer::{DocType, JsonTokenizer, LanceTokenizer};
@@ -37,7 +39,6 @@ use lance_index::scalar::inverted::tokenizer::lance_tokenizer::TextTokenizer;
 use lance_index::scalar::inverted::{
     FTS_SCHEMA, InvertedIndex, SCORE_COL, flat_bm25_search_stream,
 };
-use lance_index::{DatasetIndexExt, IndexCriteria};
 use lance_index::{prefilter::PreFilter, scalar::inverted::query::BooleanQuery};
 use tracing::instrument;
 
@@ -1384,6 +1385,7 @@ impl ExecutionPlan for BooleanQueryExec {
 pub mod tests {
     use std::sync::{Arc, Mutex};
 
+    use crate::index::DatasetIndexExt;
     use datafusion::{execution::TaskContext, physical_plan::ExecutionPlan};
     use lance_datafusion::datagen::DatafusionDatagenExt;
     use lance_datafusion::exec::{ExecutionStatsCallback, ExecutionSummaryCounts};
@@ -1396,7 +1398,7 @@ pub mod tests {
         PhraseQuery,
     };
     use lance_index::scalar::{FullTextSearchQuery, InvertedIndexParams};
-    use lance_index::{DatasetIndexExt, IndexCriteria, IndexType};
+    use lance_index::{IndexCriteria, IndexType};
 
     use crate::{
         index::DatasetIndexInternalExt,
diff --git a/rust/lance/src/io/exec/knn.rs b/rust/lance/src/io/exec/knn.rs
index ab7589ce1a8..83240b19741 100644
--- a/rust/lance/src/io/exec/knn.rs
+++ b/rust/lance/src/io/exec/knn.rs
@@ -1344,6 +1344,7 @@ impl ExecutionPlan for MultivectorScoringExec {
 mod tests {
     use super::*;
 
+    use crate::index::DatasetIndexExt;
     use arrow::compute::{concat_batches, sort_to_indices, take_record_batch};
     use arrow::datatypes::Float32Type;
     use arrow_array::{
@@ -1353,10 +1354,10 @@ mod tests {
     use lance_core::utils::tempfile::TempStrDir;
     use lance_datafusion::exec::{ExecutionStatsCallback, ExecutionSummaryCounts};
     use lance_datagen::{BatchCount, RowCount, array};
+    use lance_index::IndexType;
     use lance_index::optimize::OptimizeOptions;
     use lance_index::vector::ivf::IvfBuildParams;
     use lance_index::vector::pq::PQBuildParams;
-    use lance_index::{DatasetIndexExt, IndexType};
     use lance_linalg::distance::MetricType;
     use lance_testing::datagen::generate_random_array;
     use rstest::rstest;
diff --git a/rust/lance/src/io/exec/scalar_index.rs b/rust/lance/src/io/exec/scalar_index.rs
index aafd3c8d2eb..f587ec22a91 100644
--- a/rust/lance/src/io/exec/scalar_index.rs
+++ b/rust/lance/src/io/exec/scalar_index.rs
@@ -7,7 +7,7 @@ use super::utils::{IndexMetrics, InstrumentedRecordBatchStreamAdapter};
 use crate::{
     Dataset,
     dataset::rowids::load_row_id_sequences,
-    index::{DatasetIndexInternalExt, prefilter::DatasetPreFilter},
+    index::{DatasetIndexExt, DatasetIndexInternalExt, prefilter::DatasetPreFilter},
 };
 use arrow_array::{Array, RecordBatch, UInt64Array};
 use arrow_schema::{Schema, SchemaRef};
@@ -40,7 +40,7 @@ use lance_datafusion::{
     },
 };
 use lance_index::{
-    DatasetIndexExt, IndexCriteria,
+    IndexCriteria,
     metrics::MetricsCollector,
     scalar::{
         SargableQuery, ScalarIndex,
@@ -733,6 +733,7 @@ impl ExecutionPlan for MaterializeIndexExec {
 mod tests {
     use std::{ops::Bound, sync::Arc};
 
+    use crate::index::DatasetIndexExt;
     use arrow::datatypes::UInt64Type;
     use datafusion::{
         execution::TaskContext, physical_plan::ExecutionPlan, prelude::SessionConfig,
@@ -742,7 +743,7 @@ mod tests {
     use lance_core::utils::tempfile::TempStrDir;
     use lance_datagen::gen_batch;
     use lance_index::{
-        DatasetIndexExt, IndexType,
+        IndexType,
         scalar::{
             SargableQuery, ScalarIndexParams,
             expression::{ScalarIndexExpr, ScalarIndexSearch},
diff --git a/rust/lance/src/session/index_extension.rs b/rust/lance/src/session/index_extension.rs
index b500d52086c..2055f64e340 100644
--- a/rust/lance/src/session/index_extension.rs
+++ b/rust/lance/src/session/index_extension.rs
@@ -65,6 +65,7 @@ mod test {
         sync::{Arc, atomic::AtomicBool},
     };
 
+    use crate::index::DatasetIndexExt;
     use arrow_array::{Float32Array, RecordBatch, UInt32Array};
     use arrow_schema::Schema;
     use datafusion::execution::SendableRecordBatchStream;
@@ -75,8 +76,7 @@ mod test {
     use lance_file::version::LanceFileVersion;
     use lance_index::vector::v3::subindex::SubIndexType;
     use lance_index::{
-        DatasetIndexExt, INDEX_FILE_NAME, INDEX_METADATA_SCHEMA_KEY, Index, IndexMetadata,
-        IndexType,
+        INDEX_FILE_NAME, INDEX_METADATA_SCHEMA_KEY, Index, IndexMetadata, IndexType,
         vector::{Query, hnsw::VECTOR_ID_FIELD},
     };
     use lance_index::{
diff --git a/rust/lance/tests/query/inverted.rs b/rust/lance/tests/query/inverted.rs
index 63b36b6a823..c9ce1231d92 100644
--- a/rust/lance/tests/query/inverted.rs
+++ b/rust/lance/tests/query/inverted.rs
@@ -7,9 +7,10 @@ use arrow_array::{ArrayRef, Int32Array, RecordBatch, StringArray, UInt32Array};
 use lance::Dataset;
 use lance::dataset::scanner::ColumnOrdering;
 use lance::dataset::{InsertBuilder, WriteParams};
+use lance::index::DatasetIndexExt;
+use lance_index::IndexType;
 use lance_index::scalar::inverted::query::{FtsQuery, PhraseQuery};
 use lance_index::scalar::{FullTextSearchQuery, InvertedIndexParams};
-use lance_index::{DatasetIndexExt, IndexType};
 use tantivy::tokenizer::Language;
 
 use super::{strip_score_column, test_fts, test_scan, test_take};
diff --git a/rust/lance/tests/query/primitives.rs b/rust/lance/tests/query/primitives.rs
index 6c72f66bd23..65fa6f4e4d3 100644
--- a/rust/lance/tests/query/primitives.rs
+++ b/rust/lance/tests/query/primitives.rs
@@ -13,8 +13,9 @@ use lance::Dataset;
 use lance::dataset::WriteParams;
 use lance::dataset::optimize::{CompactionOptions, compact_files};
 
+use lance::index::DatasetIndexExt;
 use lance_datagen::{ArrayGeneratorExt, RowCount, array, gen_batch};
-use lance_index::{DatasetIndexExt, IndexType};
+use lance_index::IndexType;
 
 use super::{test_filter, test_scan, test_take};
 use crate::utils::DatasetTestCases;
diff --git a/rust/lance/tests/utils/mod.rs b/rust/lance/tests/utils/mod.rs
index c407c62c03b..b8a034a50b0 100644
--- a/rust/lance/tests/utils/mod.rs
+++ b/rust/lance/tests/utils/mod.rs
@@ -7,6 +7,7 @@ use std::sync::Arc;
 
 use arrow_array::{ArrayRef, Int32Array, RecordBatch};
 use futures::FutureExt;
+use lance::index::DatasetIndexExt;
 use lance::index::vector::VectorIndexParams;
 use lance::{
     Dataset,
@@ -17,7 +18,7 @@ use lance_index::vector::hnsw::builder::HnswBuildParams;
 use lance_index::vector::ivf::IvfBuildParams;
 use lance_index::vector::pq::PQBuildParams;
 use lance_index::vector::sq::builder::SQBuildParams;
-use lance_index::{DatasetIndexExt, IndexParams, IndexType};
+use lance_index::{IndexParams, IndexType};
 use lance_linalg::distance::{DistanceType, MetricType};
 
 #[derive(Clone, Copy, Debug)]