diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 439d400..0bab98e 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -10,6 +10,7 @@ repos:
     additional_dependencies:
       - 'daft>=0.7.0'
       - 'pylance>=6.0.0'
+      - 'pyarrow-stubs>=20.0.0'
 
 - repo: https://github.com/pre-commit/pre-commit-hooks
   rev: v6.0.0
@@ -26,6 +27,7 @@ repos:
   hooks:
   - id: pretty-format-toml
     args: [--autofix, --no-sort]
+    exclude: uv.lock
 
 - repo: https://github.com/astral-sh/ruff-pre-commit
   # ruff version.
diff --git a/daft_lance/lance_merge_column.py b/daft_lance/lance_merge_column.py
index ac624b7..595bdd5 100644
--- a/daft_lance/lance_merge_column.py
+++ b/daft_lance/lance_merge_column.py
@@ -142,13 +142,13 @@ def __call__(self, *cols: Any) -> list[dict[str, bytes]]:
                 f"GroupFragmentMergeUDF expected {len(self.read_columns)} data columns, received {len(data_cols)}."
             )
 
-        arrays: list[_pa.Array] = []
+        arrays: list[_pa.Array[Any]] = []
 
         for col_name, s in zip(self.read_columns, data_cols):
             pylist = s.to_pylist() if hasattr(s, "to_pylist") else list(s)
 
             if col_name == self.right_on:
-                key_arr: _pa.Array
+                key_arr: _pa.Array[Any]
                 if self.right_on == "_rowaddr":
                     key_arr = _pa.array(pylist, type=_pa.uint64())
                 else:
diff --git a/daft_lance/lance_scan.py b/daft_lance/lance_scan.py
index de3349f..e0ea8fb 100644
--- a/daft_lance/lance_scan.py
+++ b/daft_lance/lance_scan.py
@@ -143,7 +143,10 @@ def __init__(
         self._enable_strict_filter_pushdown = get_context().daft_planning_config.enable_strict_filter_pushdown
         base = self._ds.schema
         if self._include_fragment_id:
-            base = pa.schema([*base, pa.field("fragment_id", pa.int64())], metadata=base.metadata)
+            metadata: dict[bytes | str, bytes | str] | None = (
+                {k: v for k, v in base.metadata.items()} if base.metadata else None
+            )
+            base = pa.schema([*base, pa.field("fragment_id", pa.int64())], metadata=metadata)
         self._schema = convert_lance_schema(base)
 
     def name(self) -> str:
diff --git a/daft_lance/utils.py b/daft_lance/utils.py
index 0488262..0eb32c7 100644
--- a/daft_lance/utils.py
+++ b/daft_lance/utils.py
@@ -106,14 +106,14 @@ def construct_lance_dataset(
     }
     effective_kwargs.update(kwargs or {})
     try:
-        ds._lance_open_kwargs = effective_kwargs
+        ds._lance_open_kwargs = effective_kwargs  # type: ignore[attr-defined]
     except Exception:
         pass
 
     # Preserve the full user-provided defaults (including nearest) for Daft's planning
     # even if we stripped keys out before calling `lance.dataset`.
     try:
-        ds._daft_default_scan_options = original_default_scan_options
+        ds._daft_default_scan_options = original_default_scan_options  # type: ignore[attr-defined]
     except Exception:
         pass
 
@@ -174,4 +174,7 @@ def select_required_columns(schema: pa.Schema, required_columns: list[str] | Non
     if missing:
         raise KeyError(f"Required columns missing in schema: {missing}")
     fields = [schema.field(schema.get_field_index(c)) for c in required_columns]
-    return pa.schema(fields, metadata=schema.metadata)
+    metadata: dict[bytes | str, bytes | str] | None = (
+        {k: v for k, v in schema.metadata.items()} if schema.metadata else None
+    )
+    return pa.schema(fields, metadata=metadata)