diff --git a/ml_engine/custom-prediction-routines/README.md b/ml_engine/custom-prediction-routines/README.md
new file mode 100644
index 00000000000..86e66e8e2cb
--- /dev/null
+++ b/ml_engine/custom-prediction-routines/README.md
@@ -0,0 +1,28 @@
+# Custom prediction routines (beta)
+
+Read the AI Platform documentation about custom prediction routines to learn how
+to use these samples:
+
+* [Custom prediction routines (with a TensorFlow Keras
+  example)](https://cloud.google.com/ml-engine/docs/tensorflow/custom-prediction-routines)
+* [Custom prediction routines (with a scikit-learn
+  example)](https://cloud.google.com/ml-engine/docs/scikit/custom-prediction-routines)
+
+If you want to package a predictor directly from this directory, make sure to
+edit `setup.py`: replace the reference to `predictor.py` with either
+`tensorflow-predictor.py` or `scikit-predictor.py`.
+
+## What's next
+
+For a more complete example of how to train and deploy a custom prediction
+routine, check out one of the following tutorials:
+
+* [Creating a custom prediction routine with
+  Keras](https://cloud.google.com/ml-engine/docs/tensorflow/custom-prediction-routine-keras)
+  (also available as [a Jupyter
+  notebook](https://colab.research.google.com/github/GoogleCloudPlatform/cloudml-samples/blob/master/notebooks/tensorflow/custom-prediction-routine-keras.ipynb))
+
+* [Creating a custom prediction routine with
+  scikit-learn](https://cloud.google.com/ml-engine/docs/scikit/custom-prediction-routine-scikit-learn)
+  (also available as [a Jupyter
+  notebook](https://colab.research.google.com/github/GoogleCloudPlatform/cloudml-samples/blob/master/notebooks/scikit-learn/custom-prediction-routine-scikit-learn.ipynb))
\ No newline at end of file
diff --git a/ml_engine/custom-prediction-routines/predictor-interface.py b/ml_engine/custom-prediction-routines/predictor-interface.py
new file mode 100644
index 00000000000..a45ea763f80
--- /dev/null
+++ b/ml_engine/custom-prediction-routines/predictor-interface.py
@@ -0,0 +1,50 @@
+# Copyright 2019 Google LLC
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+
+#     https://www.apache.org/licenses/LICENSE-2.0
+
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+class Predictor(object):
+    """Interface for constructing custom predictors."""
+
+    def predict(self, instances, **kwargs):
+        """Performs custom prediction.
+
+        Instances are the decoded values from the request. They have already
+        been deserialized from JSON.
+
+        Args:
+            instances: A list of prediction input instances.
+            **kwargs: A dictionary of keyword args provided as additional
+                fields on the predict request body.
+
+        Returns:
+            A list of outputs containing the prediction results. This list must
+            be JSON serializable.
+        """
+        raise NotImplementedError()
+
+    @classmethod
+    def from_path(cls, model_dir):
+        """Creates an instance of Predictor using the given path.
+
+        Loading of the predictor should be done in this method.
+
+        Args:
+            model_dir: The local directory that contains the exported model
+                file along with any additional files uploaded when creating the
+                version resource.
+
+        Returns:
+            An instance implementing this Predictor class.
+        """
+        raise NotImplementedError()
diff --git a/ml_engine/custom-prediction-routines/preprocess.py b/ml_engine/custom-prediction-routines/preprocess.py
new file mode 100644
index 00000000000..e28aaf357df
--- /dev/null
+++ b/ml_engine/custom-prediction-routines/preprocess.py
@@ -0,0 +1,43 @@
+# Copyright 2019 Google LLC
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+
+#     https://www.apache.org/licenses/LICENSE-2.0
+
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+
+
+class ZeroCenterer(object):
+    """Stores means of each column of a matrix and uses them for preprocessing.
+    """
+
+    def __init__(self):
+        """On initialization, is not tied to any distribution."""
+        self._means = None
+
+    def preprocess(self, data):
+        """Transforms a matrix.
+
+        The first time this is called, it stores the means of each column of
+        the input. Then it transforms the input so each column has mean 0. For
+        subsequent calls, it subtracts the stored means from each column. This
+        lets you 'center' data at prediction time based on the distribution of
+        the original training data.
+
+        Args:
+            data: A NumPy matrix of numerical data.
+
+        Returns:
+            A transformed matrix with the same dimensions as the input.
+        """
+        if self._means is None:  # during training only
+            self._means = np.mean(data, axis=0)
+        return data - self._means
diff --git a/ml_engine/custom-prediction-routines/scikit-predictor.py b/ml_engine/custom-prediction-routines/scikit-predictor.py
new file mode 100644
index 00000000000..ca2998bc68f
--- /dev/null
+++ b/ml_engine/custom-prediction-routines/scikit-predictor.py
@@ -0,0 +1,73 @@
+# Copyright 2019 Google LLC
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+
+#     https://www.apache.org/licenses/LICENSE-2.0
+
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import pickle
+
+import numpy as np
+from sklearn.externals import joblib
+
+
+class MyPredictor(object):
+    """An example Predictor for an AI Platform custom prediction routine."""
+
+    def __init__(self, model, preprocessor):
+        """Stores artifacts for prediction. Only initialized via `from_path`.
+        """
+        self._model = model
+        self._preprocessor = preprocessor
+
+    def predict(self, instances, **kwargs):
+        """Performs custom prediction.
+
+        Preprocesses inputs, then performs prediction using the trained
+        scikit-learn model.
+
+        Args:
+            instances: A list of prediction input instances.
+            **kwargs: A dictionary of keyword args provided as additional
+                fields on the predict request body.
+
+        Returns:
+            A list of outputs containing the prediction results.
+        """
+        inputs = np.asarray(instances)
+        preprocessed_inputs = self._preprocessor.preprocess(inputs)
+        outputs = self._model.predict(preprocessed_inputs)
+        return outputs.tolist()
+
+    @classmethod
+    def from_path(cls, model_dir):
+        """Creates an instance of MyPredictor using the given path.
+
+        This loads artifacts that have been copied from your model directory in
+        Cloud Storage. MyPredictor uses them during prediction.
+
+        Args:
+            model_dir: The local directory that contains the trained
+                scikit-learn model and the pickled preprocessor instance. These
+                are copied from the Cloud Storage model directory you provide
+                when you deploy a version resource.
+
+        Returns:
+            An instance of `MyPredictor`.
+        """
+        model_path = os.path.join(model_dir, 'model.joblib')
+        model = joblib.load(model_path)
+
+        preprocessor_path = os.path.join(model_dir, 'preprocessor.pkl')
+        with open(preprocessor_path, 'rb') as f:
+            preprocessor = pickle.load(f)
+
+        return cls(model, preprocessor)
diff --git a/ml_engine/custom-prediction-routines/setup.py b/ml_engine/custom-prediction-routines/setup.py
new file mode 100644
index 00000000000..f313c0f1404
--- /dev/null
+++ b/ml_engine/custom-prediction-routines/setup.py
@@ -0,0 +1,20 @@
+# Copyright 2019 Google LLC
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+
+#     https://www.apache.org/licenses/LICENSE-2.0
+
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from setuptools import setup
+
+setup(
+    name='my_custom_code',
+    version='0.1',
+    scripts=['predictor.py', 'preprocess.py'])
diff --git a/ml_engine/custom-prediction-routines/tensorflow-predictor.py b/ml_engine/custom-prediction-routines/tensorflow-predictor.py
new file mode 100644
index 00000000000..3d8ed8422f8
--- /dev/null
+++ b/ml_engine/custom-prediction-routines/tensorflow-predictor.py
@@ -0,0 +1,73 @@
+# Copyright 2019 Google LLC
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+
+#     https://www.apache.org/licenses/LICENSE-2.0
+
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import pickle
+
+import numpy as np
+from tensorflow import keras
+
+
+class MyPredictor(object):
+    """An example Predictor for an AI Platform custom prediction routine."""
+
+    def __init__(self, model, preprocessor):
+        """Stores artifacts for prediction. Only initialized via `from_path`.
+        """
+        self._model = model
+        self._preprocessor = preprocessor
+
+    def predict(self, instances, **kwargs):
+        """Performs custom prediction.
+
+        Preprocesses inputs, then performs prediction using the trained Keras
+        model.
+
+        Args:
+            instances: A list of prediction input instances.
+            **kwargs: A dictionary of keyword args provided as additional
+                fields on the predict request body.
+
+        Returns:
+            A list of outputs containing the prediction results.
+        """
+        inputs = np.asarray(instances)
+        preprocessed_inputs = self._preprocessor.preprocess(inputs)
+        outputs = self._model.predict(preprocessed_inputs)
+        return outputs.tolist()
+
+    @classmethod
+    def from_path(cls, model_dir):
+        """Creates an instance of MyPredictor using the given path.
+
+        This loads artifacts that have been copied from your model directory in
+        Cloud Storage. MyPredictor uses them during prediction.
+
+        Args:
+            model_dir: The local directory that contains the trained Keras
+                model and the pickled preprocessor instance. These are copied
+                from the Cloud Storage model directory you provide when you
+                deploy a version resource.
+
+        Returns:
+            An instance of `MyPredictor`.
+        """
+        model_path = os.path.join(model_dir, 'model.h5')
+        model = keras.models.load_model(model_path)
+
+        preprocessor_path = os.path.join(model_dir, 'preprocessor.pkl')
+        with open(preprocessor_path, 'rb') as f:
+            preprocessor = pickle.load(f)
+
+        return cls(model, preprocessor)