diff --git a/backend/eval_config.py b/backend/eval_config.py
deleted file mode 100644
index 62a3b8f..0000000
--- a/backend/eval_config.py
+++ /dev/null
@@ -1 +0,0 @@
-EVALS_DIR = "./evals"
diff --git a/backend/evals/__init__.py b/backend/evals/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/backend/evals/config.py b/backend/evals/config.py
new file mode 100644
index 0000000..7643027
--- /dev/null
+++ b/backend/evals/config.py
@@ -0,0 +1 @@
+EVALS_DIR = "./evals_data"
diff --git a/backend/evals/core.py b/backend/evals/core.py
new file mode 100644
index 0000000..61db1a3
--- /dev/null
+++ b/backend/evals/core.py
@@ -0,0 +1,29 @@
+import os
+
+from llm import stream_openai_response
+from prompts import assemble_prompt
+from prompts.types import Stack
+from utils import pprint_prompt
+
+
+async def generate_code_core(image_url: str, stack: Stack) -> str:
+    prompt_messages = assemble_prompt(image_url, stack)
+    openai_api_key = os.environ.get("OPENAI_API_KEY")
+    openai_base_url = None
+
+    pprint_prompt(prompt_messages)
+
+    async def process_chunk(content: str):
+        pass
+
+    if not openai_api_key:
+        raise Exception("OpenAI API key not found")
+
+    completion = await stream_openai_response(
+        prompt_messages,
+        api_key=openai_api_key,
+        base_url=openai_base_url,
+        callback=lambda x: process_chunk(x),
+    )
+
+    return completion
diff --git a/backend/eval_utils.py b/backend/evals/utils.py
similarity index 100%
rename from backend/eval_utils.py
rename to backend/evals/utils.py
diff --git a/backend/routes/evals.py b/backend/routes/evals.py
index 48a3a95..798a9d8 100644
--- a/backend/routes/evals.py
+++ b/backend/routes/evals.py
@@ -1,8 +1,8 @@
 import os
 from fastapi import APIRouter
 from pydantic import BaseModel
-from eval_utils import image_to_data_url
-from eval_config import EVALS_DIR
+from evals.utils import image_to_data_url
+from evals.config import EVALS_DIR
 
 
 router = APIRouter()
diff --git a/backend/eval.py b/backend/run_evals.py
similarity index 57%
rename from backend/eval.py
rename to backend/run_evals.py
index 75641a3..ddb7eaa 100644
--- a/backend/eval.py
+++ b/backend/run_evals.py
@@ -1,42 +1,15 @@
 # Load environment variables first
-from typing import Any, Coroutine
 from dotenv import load_dotenv
 
-from eval_config import EVALS_DIR
-from eval_utils import image_to_data_url
-from prompts.types import Stack
-
 load_dotenv()
 
 import os
-from llm import stream_openai_response
-from prompts import assemble_prompt
+from typing import Any, Coroutine
 import asyncio
 
-from utils import pprint_prompt
-
-
-async def generate_code_core(image_url: str, stack: Stack) -> str:
-    prompt_messages = assemble_prompt(image_url, stack)
-    openai_api_key = os.environ.get("OPENAI_API_KEY")
-    openai_base_url = None
-
-    pprint_prompt(prompt_messages)
-
-    async def process_chunk(content: str):
-        pass
-
-    if not openai_api_key:
-        raise Exception("OpenAI API key not found")
-
-    completion = await stream_openai_response(
-        prompt_messages,
-        api_key=openai_api_key,
-        base_url=openai_base_url,
-        callback=lambda x: process_chunk(x),
-    )
-
-    return completion
+from evals.config import EVALS_DIR
+from evals.core import generate_code_core
+from evals.utils import image_to_data_url
 
 
 async def main():