diff --git a/surfsense_backend/alembic/versions/104_add_chunks_document_id_index.py b/surfsense_backend/alembic/versions/104_add_chunks_document_id_index.py
new file mode 100644
index 000000000..4e6dc6a15
--- /dev/null
+++ b/surfsense_backend/alembic/versions/104_add_chunks_document_id_index.py
@@ -0,0 +1,41 @@
+"""104_add_chunks_document_id_index
+
+Revision ID: 104
+Revises: 103
+Create Date: 2026-03-09
+
+Adds a B-tree index on chunks.document_id to speed up chunk lookups
+during hybrid search (both retrievers fetch chunks by document_id
+after RRF ranking selects the top documents).
+"""
+
+from __future__ import annotations
+
+from collections.abc import Sequence
+
+from alembic import op
+
+revision: str = "104"
+down_revision: str | None = "103"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def upgrade() -> None:
+    op.execute(
+        """
+        DO $$
+        BEGIN
+            IF NOT EXISTS (
+                SELECT 1 FROM pg_indexes
+                WHERE tablename = 'chunks' AND indexname = 'ix_chunks_document_id'
+            ) THEN
+                CREATE INDEX ix_chunks_document_id ON chunks(document_id);
+            END IF;
+        END$$;
+        """
+    )
+
+
+def downgrade() -> None:
+    op.execute("DROP INDEX IF EXISTS ix_chunks_document_id")
diff --git a/surfsense_backend/app/db.py b/surfsense_backend/app/db.py
index 9f0af4fc5..bc3d1c514 100644
--- a/surfsense_backend/app/db.py
+++ b/surfsense_backend/app/db.py
@@ -959,7 +959,7 @@ class Chunk(BaseModel, TimestampMixin):
     embedding = Column(Vector(config.embedding_model_instance.dimension))
 
     document_id = Column(
-        Integer, ForeignKey("documents.id", ondelete="CASCADE"), nullable=False
+        Integer, ForeignKey("documents.id", ondelete="CASCADE"), nullable=False, index=True
     )
     document = relationship("Document", back_populates="chunks")