diff --git a/backend/app/api/routes/chat.py b/backend/app/api/routes/chat.py
index 0f76a7a..72aa96a 100644
--- a/backend/app/api/routes/chat.py
+++ b/backend/app/api/routes/chat.py
@@ -6,8 +6,6 @@
 from pydantic import BaseModel, Field
 from sse_starlette.sse import EventSourceResponse
 
-from app.services.llm import get_english_translation, get_response_stream_async, generate_chat_title_async
-from app.services.embedding import embedding_service
 from app.services.database import get_client
 from app.core.rate_limit import limiter
 
@@ -38,6 +36,9 @@ async def generate_chat_events(request: Request, query: str, history: List[Histo
     Generator function that streams SSE events.
     It yields 'metadata' first, then chunks of 'content'.
     """
+    from app.services.llm import get_english_translation, get_response_stream_async
+    from app.services.embedding import embedding_service
+    
     # 1. Translate Korean query to English // Note: We don't translate history here to save costs and reduce latency
     try:
         english_query = await asyncio.wait_for(
@@ -153,6 +154,8 @@ async def chat_title_endpoint(request: Request, title_request: TitleRequest):
     """
     Endpoint for generating a short chat room title based on the first user query.
     """
+    from app.services.llm import generate_chat_title_async
+    
     query = title_request.query.strip()
     if not query:
         return {"title": DEFAULT_CHAT_TITLE}
diff --git a/backend/app/main.py b/backend/app/main.py
index 9eb9212..4cf370f 100644
--- a/backend/app/main.py
+++ b/backend/app/main.py
@@ -6,6 +6,7 @@
 
 from app.api.routes import chat
 from app.core.rate_limit import limiter
+import asyncio
 from contextlib import asynccontextmanager
 import logging
 
@@ -48,6 +49,8 @@ def _on_preload_done(task: asyncio.Task):
                 await asyncio.wait_for(asyncio.shield(preload_task), timeout=3.0)
             except asyncio.TimeoutError:
                 logger.warning("Preload task did not finish before shutdown.")
+            except Exception as e:
+                logger.exception("Exception occurred while waiting for preload task during shutdown.")
 
 app = FastAPI(
     title="PhiloRAG API",
@@ -83,10 +86,13 @@ async def readiness_check():
         return JSONResponse({"status": "not_ready"}, status_code=503)
         
     if preload_task.cancelled():
+        logger.warning("Preload task was cancelled during readiness check")
         return JSONResponse({"status": "failed"}, status_code=503)
         
     try:
         preload_task.result()  # re-raises if failed
-        return {"status": "ready"}
-    except Exception:
+    except Exception as e:
+        logger.warning("Preload task failed during readiness check: %s", e)
         return JSONResponse({"status": "failed"}, status_code=503)
+    else:
+        return {"status": "ready"}
diff --git a/backend/tests/e2e/test_chat_endpoint.py b/backend/tests/e2e/test_chat_endpoint.py
index f67d651..6ff131f 100644
--- a/backend/tests/e2e/test_chat_endpoint.py
+++ b/backend/tests/e2e/test_chat_endpoint.py
@@ -10,10 +10,10 @@ def test_health_check():
     assert response.status_code == 200
     assert response.json() == {"status": "healthy"}
 
-@patch("app.api.routes.chat.embedding_service.agenerate_embedding")
+@patch("app.services.embedding.EmbeddingService.agenerate_embedding")
 @patch("app.api.routes.chat._search_documents")
-@patch("app.api.routes.chat.get_english_translation")
-@patch("app.api.routes.chat.get_response_stream_async")
+@patch("app.services.llm.get_english_translation")
+@patch("app.services.llm.get_response_stream_async")
 def test_chat_endpoint_success(mock_stream, mock_translate, mock_search, mock_embed):
     # Setup mocks
     mock_translate.return_value = "What is life?"
diff --git a/backend/tests/integration/test_supabase_match.py b/backend/tests/integration/test_supabase_match.py
index df1832f..165a4d4 100644
--- a/backend/tests/integration/test_supabase_match.py
+++ b/backend/tests/integration/test_supabase_match.py
@@ -9,10 +9,10 @@
 @pytest.mark.asyncio
 async def test_supabase_match_integration():
     # 1. We mock the embedding service to return a dummy vector
-    with patch("app.api.routes.chat.embedding_service.agenerate_embedding") as mock_embed, \
+    with patch("app.services.embedding.EmbeddingService.agenerate_embedding") as mock_embed, \
          patch("app.api.routes.chat._search_documents") as mock_search, \
-         patch("app.api.routes.chat.get_english_translation") as mock_translate, \
-         patch("app.api.routes.chat.get_response_stream_async") as mock_stream:
+         patch("app.services.llm.get_english_translation") as mock_translate, \
+         patch("app.services.llm.get_response_stream_async") as mock_stream:
          
         mock_translate.return_value = "English Question"
         mock_embed.return_value = [0.1, 0.2, 0.3]