diff --git a/vllm/__init__.py b/vllm/__init__.py
index 5ca4680227598..ca454efd44b24 100644
--- a/vllm/__init__.py
+++ b/vllm/__init__.py
@@ -3,8 +3,8 @@
 from vllm.engine.arg_utils import AsyncEngineArgs, EngineArgs
 from vllm.engine.async_llm_engine import AsyncLLMEngine
 from vllm.engine.llm_engine import LLMEngine
-from vllm.engine.ray_utils import initialize_ray_cluster
 from vllm.entrypoints.llm import LLM
+from vllm.executor.ray_utils import initialize_ray_cluster
 from vllm.model_executor.models import ModelRegistry
 from vllm.outputs import CompletionOutput, RequestOutput
 from vllm.sampling_params import SamplingParams
diff --git a/vllm/engine/async_llm_engine.py b/vllm/engine/async_llm_engine.py
index 3a2f7db679358..4b007d71e9cfc 100644
--- a/vllm/engine/async_llm_engine.py
+++ b/vllm/engine/async_llm_engine.py
@@ -10,7 +10,7 @@
 from vllm.config import ModelConfig
 from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.llm_engine import LLMEngine
-from vllm.engine.ray_utils import initialize_ray_cluster, ray
+from vllm.executor.ray_utils import initialize_ray_cluster, ray
 from vllm.logger import init_logger
 from vllm.lora.request import LoRARequest
 from vllm.outputs import RequestOutput
diff --git a/vllm/engine/llm_engine.py b/vllm/engine/llm_engine.py
index 19e58fb1722cf..56c2417d6a6e6 100644
--- a/vllm/engine/llm_engine.py
+++ b/vllm/engine/llm_engine.py
@@ -15,8 +15,8 @@
     SequenceGroupOutputProcessor)
 from vllm.engine.output_processor.stop_checker import StopChecker
 from vllm.engine.output_processor.util import create_output_by_sequence_group
-from vllm.engine.ray_utils import initialize_ray_cluster
 from vllm.executor.executor_base import ExecutorBase
+from vllm.executor.ray_utils import initialize_ray_cluster
 from vllm.logger import init_logger
 from vllm.lora.request import LoRARequest
 from vllm.outputs import RequestOutput
diff --git a/vllm/executor/ray_gpu_executor.py b/vllm/executor/ray_gpu_executor.py
index e69f104e7d5a4..14b3f803782c6 100644
--- a/vllm/executor/ray_gpu_executor.py
+++ b/vllm/executor/ray_gpu_executor.py
@@ -5,8 +5,8 @@
 from itertools import islice, repeat
 from typing import TYPE_CHECKING, Any, Dict, List, Optional, Set, Tuple
 
-from vllm.engine.ray_utils import RayWorkerWrapper, ray
 from vllm.executor.executor_base import ExecutorAsyncBase, ExecutorBase
+from vllm.executor.ray_utils import RayWorkerWrapper, ray
 from vllm.logger import init_logger
 from vllm.lora.request import LoRARequest
 from vllm.sequence import SamplerOutput, SequenceGroupMetadata
@@ -74,7 +74,7 @@ def _init_workers_ray(self, placement_group: "PlacementGroup",
 
         # The driver dummy worker does not actually use any resources.
         # It holds the resource for the driver worker.
-        self.driver_dummy_worker: RayWorkerWrapper = None
+        self.driver_dummy_worker: Optional[RayWorkerWrapper] = None
         # The remaining workers are the actual ray actors.
         self.workers: List[RayWorkerWrapper] = []
 
@@ -318,6 +318,7 @@ def _run_workers(
             driver_worker_output = self.driver_worker.execute_method(
                 method, *driver_args, **driver_kwargs)
         else:
+            assert self.driver_dummy_worker is not None
             driver_worker_output = ray.get(
                 self.driver_dummy_worker.execute_method.remote(
                     method, *driver_args, **driver_kwargs))
@@ -353,8 +354,9 @@ def _compiled_ray_dag(self):
         # a dummy value for now. It will be fixed soon.
         with InputNode() as input_data:
             forward_dag = MultiOutputNode([
-                worker.execute_model_compiled_dag_remote.bind(input_data)
-                for worker in self.workers
+                worker.execute_model_compiled_dag_remote.
+                bind(  # type: ignore[attr-defined]
+                    input_data) for worker in self.workers
             ])
         return forward_dag.experimental_compile()
 
diff --git a/vllm/engine/ray_utils.py b/vllm/executor/ray_utils.py
similarity index 100%
rename from vllm/engine/ray_utils.py
rename to vllm/executor/ray_utils.py
diff --git a/vllm/transformers_utils/tokenizer_group/__init__.py b/vllm/transformers_utils/tokenizer_group/__init__.py
index 69380d67f9b94..0195c40c27f60 100644
--- a/vllm/transformers_utils/tokenizer_group/__init__.py
+++ b/vllm/transformers_utils/tokenizer_group/__init__.py
@@ -1,7 +1,7 @@
 from typing import Optional
 
 from vllm.config import TokenizerPoolConfig
-from vllm.engine.ray_utils import ray
+from vllm.executor.ray_utils import ray
 from vllm.transformers_utils.tokenizer_group.base_tokenizer_group import (
     BaseTokenizerGroup)
 from vllm.transformers_utils.tokenizer_group.tokenizer_group import (
diff --git a/vllm/transformers_utils/tokenizer_group/ray_tokenizer_group.py b/vllm/transformers_utils/tokenizer_group/ray_tokenizer_group.py
index f3cdc00564dbb..7c605416854b8 100644
--- a/vllm/transformers_utils/tokenizer_group/ray_tokenizer_group.py
+++ b/vllm/transformers_utils/tokenizer_group/ray_tokenizer_group.py
@@ -6,7 +6,7 @@
 from transformers import PreTrainedTokenizer
 
 from vllm.config import TokenizerPoolConfig
-from vllm.engine.ray_utils import ray
+from vllm.executor.ray_utils import ray
 from vllm.lora.request import LoRARequest
 from vllm.transformers_utils.tokenizer_group.base_tokenizer_group import (
     BaseTokenizerGroup)