diff --git a/src/vector_search_mcp/engine/__init__.py b/src/vector_search_mcp/engine/__init__.py
index 9dbf918..4fd5768 100644
--- a/src/vector_search_mcp/engine/__init__.py
+++ b/src/vector_search_mcp/engine/__init__.py
@@ -1,26 +1,28 @@
 from enum import StrEnum
+from functools import cache
 from typing import Literal, overload
 
 from .qdrant_engine import QdrantEngine
 
 
-class EngineType(StrEnum):
+class Backend(StrEnum):
     QDRANT = "qdrant"
     COSMOS = "cosmos"
 
 
 @overload
-def get_engine(backend: Literal[EngineType.QDRANT]) -> QdrantEngine: ...
+def get_engine(backend: Literal[Backend.QDRANT]) -> QdrantEngine: ...
 
 
 @overload
-def get_engine(backend: Literal[EngineType.COSMOS]) -> QdrantEngine: ...
+def get_engine(backend: Literal[Backend.COSMOS]) -> QdrantEngine: ...
 
 
-def get_engine(backend: EngineType):
-    if backend == EngineType.QDRANT:
+@cache
+def get_engine(backend: Backend):
+    if backend == Backend.QDRANT:
         return QdrantEngine()
-    elif backend == EngineType.COSMOS:
+    elif backend == Backend.COSMOS:
         raise NotImplementedError("Cosmos engine is not implemented yet")
     else:
         raise ValueError(f"Unknown engine type: {backend}")
diff --git a/src/vector_search_mcp/engine/base_engine.py b/src/vector_search_mcp/engine/base_engine.py
index 0cb2b3e..1c8456a 100644
--- a/src/vector_search_mcp/engine/base_engine.py
+++ b/src/vector_search_mcp/engine/base_engine.py
@@ -30,7 +30,7 @@ class BaseEngine(ABC, Generic[ResponseType, ConditionType]):
 
     async def semantic_search(
         self,
-        vector: list[float],
+        embedding: list[float],
         collection: str,
         limit: int = 10,
         conditions: list[Condition] | None = None,
@@ -38,6 +38,6 @@ class BaseEngine(ABC, Generic[ResponseType, ConditionType]):
     ) -> list[SearchRow]:
         transformed_conditions = self.transform_conditions(conditions)
         response = await self.run_similarity_query(
-            vector, collection, limit, transformed_conditions, threshold
+            embedding, collection, limit, transformed_conditions, threshold
         )
         return self.transform_response(response)
diff --git a/src/vector_search_mcp/engine/qdrant_engine.py b/src/vector_search_mcp/engine/qdrant_engine.py
index 79c8739..a288423 100644
--- a/src/vector_search_mcp/engine/qdrant_engine.py
+++ b/src/vector_search_mcp/engine/qdrant_engine.py
@@ -4,7 +4,7 @@ from typing import final, override
 from qdrant_client import AsyncQdrantClient, models
 
 from ..config import Settings
-from ..models import SearchRow, Condition, Match, MatchAny, MatchExclude
+from ..models import Condition, Match, MatchAny, MatchExclude, SearchRow
 from .base_engine import BaseEngine
 
 __all__ = ["QdrantEngine"]
diff --git a/src/vector_search_mcp/main.py b/src/vector_search_mcp/main.py
index 68692a9..a84ee6d 100644
--- a/src/vector_search_mcp/main.py
+++ b/src/vector_search_mcp/main.py
@@ -1,9 +1,9 @@
 from fastmcp import FastMCP
 
-from .engine import QdrantEngine
+from .engine import Backend, get_engine
 
 mcp = FastMCP("Vector Search MCP")
 
-engine = QdrantEngine()
+engine = get_engine(Backend.QDRANT)
 
 _ = mcp.tool(engine.semantic_search)