diff --git a/areal/experimental/inference_service/controller/workflow.py b/areal/experimental/inference_service/controller/workflow.py
index c823e3419..98be7bb1b 100644
--- a/areal/experimental/inference_service/controller/workflow.py
+++ b/areal/experimental/inference_service/controller/workflow.py
@@ -11,6 +11,7 @@
 
 from areal.api.workflow_api import RolloutWorkflow
 from areal.infra import workflow_context
+from areal.infra.rpc.rtensor import RTensor
 from areal.infra.rpc.serialization import deserialize_value
 from areal.infra.utils.http import async_http_retry
 from areal.utils import logging, stats_tracker
@@ -245,8 +246,12 @@ async def _run_online(
         if not traj:
             return None
 
-        if "rewards" in traj and len(traj["rewards"]) > 0:
-            last_reward = float(traj["rewards"][-1])
+        rewards_tensor = traj.get("rewards")
+        if isinstance(rewards_tensor, RTensor):
+            rewards_tensor = rewards_tensor.to_local()
+
+        if rewards_tensor is not None and len(rewards_tensor) > 0:
+            last_reward = float(rewards_tensor[-1])
         elif (
             "interactions" in traj
             and traj["interactions"]
diff --git a/areal/infra/workflow_executor.py b/areal/infra/workflow_executor.py
index 949d2af0b..dcf1eabd1 100644
--- a/areal/infra/workflow_executor.py
+++ b/areal/infra/workflow_executor.py
@@ -22,6 +22,7 @@
 
 from areal.api.cli_args import InferenceEngineConfig
 from areal.api import RolloutWorkflow
+from areal.infra.rpc.rtensor import RTensor
 from .async_task_runner import (
     AsyncTaskRunner,
     TaskQueueFullError,
@@ -840,6 +841,8 @@ async def _dump_trajectory(
         if traj is None:
             return False, "trajectory is None"
 
+        traj = RTensor.localize(traj)
+
         dump_dir = self._get_dump_dir(is_eval)
         if dump_dir is None:
             return False, "dump dir is empty"