diff --git a/backend/.pre-commit-config.yaml b/backend/.pre-commit-config.yaml
index ce8a3ed..0f82051 100644
--- a/backend/.pre-commit-config.yaml
+++ b/backend/.pre-commit-config.yaml
@@ -20,6 +20,7 @@ repos:
       - id: ruff-format
 
   - repo: https://github.com/astral-sh/ty-pre-commit
-    rev: v0.0.29
+    rev: v0.0.49
     hooks:
       - id: ty
+        args: ["--project", "backend/"]
diff --git a/backend/app/core/config.py b/backend/app/core/config.py
index d8feca7..2293082 100644
--- a/backend/app/core/config.py
+++ b/backend/app/core/config.py
@@ -9,10 +9,14 @@ from pydantic_settings import BaseSettings, SettingsConfigDict
 
 
 def find_env_file() -> Path | None:
-    """Find .env file in current or parent directories."""
+    """Find .env file in current, parent, or backend/ subdirectory."""
     current = Path.cwd()
-    for path in [current, current.parent]:
-        env_file = path / ".env"
+    for candidate in [
+        current,
+        current.parent,
+        current / "backend",
+    ]:
+        env_file = candidate / ".env"
         if env_file.exists():
             return env_file
     return None
diff --git a/backend/app/main.py b/backend/app/main.py
index 71f3f47..04c96af 100644
--- a/backend/app/main.py
+++ b/backend/app/main.py
@@ -7,6 +7,8 @@ from typing import TypedDict
 
 logger = logging.getLogger(__name__)
 
+from typing import TYPE_CHECKING
+
 from fastapi import FastAPI
 from fastapi_pagination import add_pagination
 
@@ -14,6 +16,9 @@ from app.api.exception_handlers import register_exception_handlers
 from app.api.router import api_router
 from app.clients.redis import RedisClient
 from app.core.config import settings
+
+if TYPE_CHECKING:
+    from app.services.rag.reranker import RerankService
 from app.core.logfire_setup import instrument_app, setup_logfire
 from app.core.logging import setup_logging
 from app.core.middleware import RequestIDMiddleware
@@ -27,6 +32,7 @@ class LifespanState(TypedDict, total=False):
     redis: RedisClient
     embedding_service: EmbeddingService
     vector_store: BaseVectorStore
+    rerank_service: "RerankService"
 
 
 @asynccontextmanager
diff --git a/backend/app/services/rag/embeddings.py b/backend/app/services/rag/embeddings.py
index 22480f9..2b94554 100644
--- a/backend/app/services/rag/embeddings.py
+++ b/backend/app/services/rag/embeddings.py
@@ -48,14 +48,19 @@ class OpenAIEmbeddingProvider(BaseEmbeddingProvider):
     Uses OpenAI's embedding models to generate text embeddings.
     """
 
-    def __init__(self, model: str) -> None:
+    def __init__(self, model: str, base_url: str | None = None) -> None:
         """Initialize the OpenAI embedding provider.
 
         Args:
             model: The OpenAI embedding model name (e.g., 'text-embedding-3-small').
+            base_url: Explicit base URL for the OpenAI API. If not provided,
+                defaults to the standard OpenAI API, explicitly ignoring any
+                OPENAI_BASE_URL env var (which may point to a chat-only local
+                server that doesn't support embeddings).
         """
         self.model = model
-        self.client = OpenAI()
+        # Explicitly pass base_url to avoid picking up OPENAI_BASE_URL from env
+        self.client = OpenAI(base_url=base_url or "https://api.openai.com/v1")
 
     def embed_queries(self, texts: list[str]) -> list[list[float]]:
         """Embed a list of query texts using OpenAI.