diff --git a/python/orca/dev/example/run-example-tests.sh b/python/orca/dev/example/run-example-tests.sh
index fb813e8e68e..02a26d302ae 100644
--- a/python/orca/dev/example/run-example-tests.sh
+++ b/python/orca/dev/example/run-example-tests.sh
@@ -237,17 +237,17 @@ else
    export PYTHONPATH=`pwd`/analytics-zoo-tensorflow-models/slim:$PYTHONPATH
  fi
 
-echo "start example test for tensorflow distributed_training train_lenet 1"
+echo "start example test for TFPark tf_optimizer train_lenet 1"
 ${SPARK_HOME}/bin/spark-submit \
     --master ${MASTER} \
     --driver-memory 200g \
     --executor-memory 200g \
     --properties-file ${ANALYTICS_ZOO_CONF} \
-    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/train_lenet.py \
+    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/train_lenet.py \
     --jars ${ANALYTICS_ZOO_JAR} \
     --conf spark.driver.extraClassPath=${ANALYTICS_ZOO_JAR} \
     --conf spark.executor.extraClassPath=${ANALYTICS_ZOO_JAR} \
-    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/train_lenet.py 1 1000\
+    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/train_lenet.py 1 1000\
 
 sed "s%/tmp%analytics-zoo-tensorflow-models%g;s%models/slim%slim%g"
 if [ -d analytics-zoo-tensorflow-models/slim ]
@@ -269,11 +269,11 @@ ${SPARK_HOME}/bin/spark-submit \
     --driver-memory 200g \
     --executor-memory 200g \
     --properties-file ${ANALYTICS_ZOO_CONF} \
-    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/evaluate_lenet.py \
+    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/evaluate_lenet.py \
     --jars ${ANALYTICS_ZOO_JAR} \
     --conf spark.driver.extraClassPath=${ANALYTICS_ZOO_JAR} \
     --conf spark.executor.extraClassPath=${ANALYTICS_ZOO_JAR} \
-    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/evaluate_lenet.py 1000\
+    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/evaluate_lenet.py 1000\
 
 echo "start example test for tensorflow distributed_training train_mnist_keras 3"
 ${SPARK_HOME}/bin/spark-submit \
@@ -281,11 +281,11 @@ ${SPARK_HOME}/bin/spark-submit \
     --driver-memory 200g \
     --executor-memory 200g \
     --properties-file ${ANALYTICS_ZOO_CONF} \
-    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/train_mnist_keras.py \
+    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/train_mnist_keras.py \
     --jars ${ANALYTICS_ZOO_JAR} \
     --conf spark.driver.extraClassPath=${ANALYTICS_ZOO_JAR} \
     --conf spark.executor.extraClassPath=${ANALYTICS_ZOO_JAR} \
-    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/train_mnist_keras.py 1 1000\
+    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/train_mnist_keras.py 1 1000\
 
 echo "start example test for tensorflow distributed_training evaluate_lenet 4"
 ${SPARK_HOME}/bin/spark-submit \
@@ -293,11 +293,11 @@ ${SPARK_HOME}/bin/spark-submit \
     --driver-memory 200g \
     --executor-memory 200g \
     --properties-file ${ANALYTICS_ZOO_CONF} \
-    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/evaluate_mnist_keras.py \
+    --py-files ${ANALYTICS_ZOO_PYZIP},${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/evaluate_mnist_keras.py \
     --jars ${ANALYTICS_ZOO_JAR} \
     --conf spark.driver.extraClassPath=${ANALYTICS_ZOO_JAR} \
     --conf spark.executor.extraClassPath=${ANALYTICS_ZOO_JAR} \
-    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/distributed_training/evaluate_mnist_keras.py 1000\
+    ${ANALYTICS_ZOO_ROOT}/pyzoo/zoo/examples/tensorflow/tfpark/tf_optimizer/evaluate_mnist_keras.py 1000\
 
     
 now=$(date "+%s")
diff --git a/python/orca/example/tfpark/README.md b/python/orca/example/tfpark/README.md
index 971624b4594..01ab215777a 100644
--- a/python/orca/example/tfpark/README.md
+++ b/python/orca/example/tfpark/README.md
@@ -41,7 +41,7 @@ Using TFDataset as data input
 export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
 export SPARK_HOME=... # the root directory of Spark
 
-sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] keras_dataset.py
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] keras/keras_dataset.py
 ```
 
 Using numpy.ndarray as data input
@@ -49,14 +49,14 @@ Using numpy.ndarray as data input
 export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
 export SPARK_HOME=... # the root directory of Spark
 
-sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] keras_ndarray.py
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] keras/keras_ndarray.py
 ```
 
 ## Run the TFEstimator example after pip install
 
 Using TFDataset as data input
 ```bash
-python estimator_dataset.py
+python estimator/estimator_dataset.py
 ```
 
 Using FeatureSet as data input
@@ -76,7 +76,7 @@ IMAGE_PATH=...
 NUM_CLASSES=..
 
 
-python estimator_inception.py --image-path $IMAGE_PATH --num-classes $NUM_CLASSES
+python estimator/estimator_inception.py --image-path $IMAGE_PATH --num-classes $NUM_CLASSES
 ```
 
 ## Run the TFEstimator example with prebuilt package
@@ -86,7 +86,7 @@ Using TFDataset as data input
 export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
 export SPARK_HOME=... # the root directory of Spark
 
-sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] estimator_dataset.py
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] estimator/estimator_dataset.py
 ```
 
 Using FeatureSet as data input
@@ -110,5 +110,65 @@ IMAGE_PATH=...
 NUM_CLASSES=..
 
 
-sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] estimator_inception.py --image-path $IMAGE_PATH --num-classes $NUM_CLASSES
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] estimator/estimator_inception.py --image-path $IMAGE_PATH --num-classes $NUM_CLASSES
+```
+
+## Run the Training Example using TFOptimizer after pip install
+
+```bash
+python tf_optimzer/train_lenet.py
+```
+
+For Keras users:
+
+```bash
+python tf_optimizer/train_mnist_keras.py
+```
+
+## Run the Training Example using TFOptimizer with prebuilt package
+
+```bash
+export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
+export SPARK_HOME=... # the root directory of Spark
+
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] tf_optimizer/train_lenet.py
+```
+
+For Keras users:
+
+```bash
+export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
+export SPARK_HOME=... # the root directory of Spark
+
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] tf_optimizer/train_mnist_keras.py
+```
+
+## Run the Evaluation Example using TFPredictor after pip install
+
+```bash
+python tf_optimizer/evaluate_lenet.py
+```
+
+For Keras users:
+
+```bash
+python tf_optimizer/evaluate_mnist_keras.py
+```
+
+## Run the Evaluation Example using TFPredictor with prebuilt package
+
+```bash
+export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
+export SPARK_HOME=... # the root directory of Spark
+
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] tf_optimizer/evaluate_lenet.py
+```
+
+For Keras users:
+
+```bash
+export ANALYTICS_ZOO_HOME=... # the directory where you extract the downloaded Analytics Zoo zip package
+export SPARK_HOME=... # the root directory of Spark
+
+sh $ANALYTICS_ZOO_HOME/bin/spark-submit-python-with-zoo.sh --master local[4] tf_optimizer/evaluate_mnist_keras.py
 ```
\ No newline at end of file
diff --git a/python/orca/example/tfpark/estimator/__init__.py b/python/orca/example/tfpark/estimator/__init__.py
new file mode 100644
index 00000000000..5976dc4df02
--- /dev/null
+++ b/python/orca/example/tfpark/estimator/__init__.py
@@ -0,0 +1,15 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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.
+#
diff --git a/python/orca/example/tfpark/estimator_dataset.py b/python/orca/example/tfpark/estimator/estimator_dataset.py
similarity index 100%
rename from python/orca/example/tfpark/estimator_dataset.py
rename to python/orca/example/tfpark/estimator/estimator_dataset.py
diff --git a/python/orca/example/tfpark/estimator_inception.py b/python/orca/example/tfpark/estimator/estimator_inception.py
similarity index 100%
rename from python/orca/example/tfpark/estimator_inception.py
rename to python/orca/example/tfpark/estimator/estimator_inception.py
diff --git a/python/orca/example/tfpark/keras/__init__.py b/python/orca/example/tfpark/keras/__init__.py
new file mode 100644
index 00000000000..5976dc4df02
--- /dev/null
+++ b/python/orca/example/tfpark/keras/__init__.py
@@ -0,0 +1,15 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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.
+#
diff --git a/python/orca/example/tfpark/keras_dataset.py b/python/orca/example/tfpark/keras/keras_dataset.py
similarity index 100%
rename from python/orca/example/tfpark/keras_dataset.py
rename to python/orca/example/tfpark/keras/keras_dataset.py
diff --git a/python/orca/example/tfpark/keras_ndarray.py b/python/orca/example/tfpark/keras/keras_ndarray.py
similarity index 100%
rename from python/orca/example/tfpark/keras_ndarray.py
rename to python/orca/example/tfpark/keras/keras_ndarray.py
diff --git a/python/orca/example/tfpark/tf_optimizer/__init__.py b/python/orca/example/tfpark/tf_optimizer/__init__.py
new file mode 100644
index 00000000000..5976dc4df02
--- /dev/null
+++ b/python/orca/example/tfpark/tf_optimizer/__init__.py
@@ -0,0 +1,15 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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.
+#
diff --git a/python/orca/example/tfpark/tf_optimizer/evaluate_lenet.py b/python/orca/example/tfpark/tf_optimizer/evaluate_lenet.py
new file mode 100644
index 00000000000..df3511ce554
--- /dev/null
+++ b/python/orca/example/tfpark/tf_optimizer/evaluate_lenet.py
@@ -0,0 +1,81 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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 tensorflow as tf
+from zoo import init_nncontext
+from zoo.tfpark import TFDataset, TFPredictor
+import numpy as np
+import sys
+
+from bigdl.dataset import mnist
+from bigdl.dataset.transformer import *
+
+sys.path.append("/tmp/models/slim")  # add the slim library
+from nets import lenet
+
+slim = tf.contrib.slim
+
+
+def main(data_num):
+
+    sc = init_nncontext()
+
+    # get data, pre-process and create TFDataset
+    (images_data, labels_data) = mnist.read_data_sets("/tmp/mnist", "test")
+    image_rdd = sc.parallelize(images_data[:data_num])
+    labels_rdd = sc.parallelize(labels_data[:data_num])
+    rdd = image_rdd.zip(labels_rdd) \
+        .map(lambda rec_tuple: [normalizer(rec_tuple[0], mnist.TRAIN_MEAN, mnist.TRAIN_STD),
+                                np.array(rec_tuple[1])])
+
+    dataset = TFDataset.from_rdd(rdd,
+                                 names=["features", "labels"],
+                                 shapes=[[28, 28, 1], [1]],
+                                 types=[tf.float32, tf.int32],
+                                 batch_per_thread=20
+                                 )
+
+    # construct the model from TFDataset
+    images, labels = dataset.tensors
+
+    labels = tf.squeeze(labels)
+
+    with slim.arg_scope(lenet.lenet_arg_scope()):
+        logits, end_points = lenet.lenet(images, num_classes=10, is_training=False)
+
+    predictions = tf.to_int32(tf.argmax(logits, axis=1))
+    correct = tf.expand_dims(tf.to_int32(tf.equal(predictions, labels)), axis=1)
+
+    saver = tf.train.Saver()
+
+    with tf.Session() as sess:
+        sess.run(tf.global_variables_initializer())
+        saver.restore(sess, "/tmp/lenet/model")
+
+        predictor = TFPredictor(sess, [correct])
+
+        accuracy = predictor.predict().mean()
+
+        print("predict accuracy is %s" % accuracy)
+
+
+if __name__ == '__main__':
+
+    data_num = 10000
+
+    if len(sys.argv) > 1:
+        data_num = int(sys.argv[1])
+    main(data_num)
diff --git a/python/orca/example/tfpark/tf_optimizer/evaluate_mnist_keras.py b/python/orca/example/tfpark/tf_optimizer/evaluate_mnist_keras.py
new file mode 100644
index 00000000000..1ec7eca6367
--- /dev/null
+++ b/python/orca/example/tfpark/tf_optimizer/evaluate_mnist_keras.py
@@ -0,0 +1,85 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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 tensorflow as tf
+from zoo import init_nncontext
+from zoo.tfpark import TFDataset, TFPredictor
+import numpy as np
+import sys
+from tensorflow.keras.models import Model
+from tensorflow.keras.layers import *
+
+from bigdl.dataset import mnist
+from bigdl.dataset.transformer import *
+
+DISTRIBUTED = True
+
+
+def main(data_num):
+
+    data = Input(shape=[28, 28, 1])
+
+    x = Flatten()(data)
+    x = Dense(64, activation='relu')(x)
+    x = Dense(64, activation='relu')(x)
+    predictions = Dense(10, activation='softmax')(x)
+
+    model = Model(inputs=data, outputs=predictions)
+
+    model.load_weights("/tmp/mnist_keras.h5")
+
+    if DISTRIBUTED:
+        # using RDD api to do distributed evaluation
+        sc = init_nncontext()
+        # get data, pre-process and create TFDataset
+        (images_data, labels_data) = mnist.read_data_sets("/tmp/mnist", "test")
+        image_rdd = sc.parallelize(images_data[:data_num])
+        labels_rdd = sc.parallelize(labels_data[:data_num])
+        rdd = image_rdd.zip(labels_rdd) \
+            .map(lambda rec_tuple: [normalizer(rec_tuple[0], mnist.TRAIN_MEAN, mnist.TRAIN_STD)])
+
+        dataset = TFDataset.from_rdd(rdd,
+                                     names=["features"],
+                                     shapes=[[28, 28, 1]],
+                                     types=[tf.float32],
+                                     batch_per_thread=20
+                                     )
+        predictor = TFPredictor.from_keras(model, dataset)
+
+        accuracy = predictor.predict().zip(labels_rdd).map(lambda x: np.argmax(x[0]) == x[1]).mean()
+
+        print("predict accuracy is %s" % accuracy)
+
+    else:
+        # using keras api for local evaluation
+        model.compile(optimizer='rmsprop',
+                      loss='sparse_categorical_crossentropy',
+                      metrics=['accuracy'])
+
+        (images_data, labels_data) = mnist.read_data_sets("/tmp/mnist", "test")
+        images_data = normalizer(images_data, mnist.TRAIN_MEAN, mnist.TRAIN_STD)
+        result = model.evaluate(images_data, labels_data)
+        print(model.metrics_names)
+        print(result)
+
+
+if __name__ == '__main__':
+
+    data_num = 10000
+
+    if len(sys.argv) > 1:
+        data_num = int(sys.argv[1])
+    main(data_num)
diff --git a/python/orca/example/tfpark/tf_optimizer/train_lenet.py b/python/orca/example/tfpark/tf_optimizer/train_lenet.py
new file mode 100644
index 00000000000..77a9e95345d
--- /dev/null
+++ b/python/orca/example/tfpark/tf_optimizer/train_lenet.py
@@ -0,0 +1,86 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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 heapq
+
+import tensorflow as tf
+from zoo import init_nncontext
+from zoo.tfpark import TFOptimizer, TFDataset
+from bigdl.optim.optimizer import *
+import numpy as np
+import sys
+
+from bigdl.dataset import mnist
+from bigdl.dataset.transformer import *
+
+sys.path.append("/tmp/models/slim")  # add the slim library
+from nets import lenet
+
+slim = tf.contrib.slim
+
+
+def main(max_epoch, data_num):
+    sc = init_nncontext()
+
+    # get data, pre-process and create TFDataset
+    def get_data_rdd(dataset):
+        (images_data, labels_data) = mnist.read_data_sets("/tmp/mnist", dataset)
+        image_rdd = sc.parallelize(images_data[:data_num])
+        labels_rdd = sc.parallelize(labels_data[:data_num])
+        rdd = image_rdd.zip(labels_rdd) \
+            .map(lambda rec_tuple: [normalizer(rec_tuple[0], mnist.TRAIN_MEAN, mnist.TRAIN_STD),
+                                    np.array(rec_tuple[1])])
+        return rdd
+
+    training_rdd = get_data_rdd("train")
+    testing_rdd = get_data_rdd("test")
+    dataset = TFDataset.from_rdd(training_rdd,
+                                 names=["features", "labels"],
+                                 shapes=[[28, 28, 1], []],
+                                 types=[tf.float32, tf.int32],
+                                 batch_size=280,
+                                 val_rdd=testing_rdd
+                                 )
+
+    # construct the model from TFDataset
+    images, labels = dataset.tensors
+
+    with slim.arg_scope(lenet.lenet_arg_scope()):
+        logits, end_points = lenet.lenet(images, num_classes=10, is_training=True)
+
+    loss = tf.reduce_mean(tf.losses.sparse_softmax_cross_entropy(logits=logits, labels=labels))
+
+    # create a optimizer
+    optimizer = TFOptimizer(loss, Adam(1e-3),
+                            val_outputs=[logits],
+                            val_labels=[labels],
+                            val_method=Top1Accuracy())
+    optimizer.set_train_summary(TrainSummary("/tmp/az_lenet", "lenet"))
+    optimizer.set_val_summary(ValidationSummary("/tmp/az_lenet", "lenet"))
+    # kick off training
+    optimizer.optimize(end_trigger=MaxEpoch(max_epoch))
+
+    saver = tf.train.Saver()
+    saver.save(optimizer.sess, "/tmp/lenet/model")
+
+if __name__ == '__main__':
+
+    max_epoch = 5
+    data_num = 60000
+
+    if len(sys.argv) > 1:
+        max_epoch = int(sys.argv[1])
+        data_num = int(sys.argv[2])
+    main(max_epoch, data_num)
diff --git a/python/orca/example/tfpark/tf_optimizer/train_mnist_keras.py b/python/orca/example/tfpark/tf_optimizer/train_mnist_keras.py
new file mode 100644
index 00000000000..de7339d0d54
--- /dev/null
+++ b/python/orca/example/tfpark/tf_optimizer/train_mnist_keras.py
@@ -0,0 +1,82 @@
+#
+# Copyright 2018 Analytics Zoo Authors.
+#
+# 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
+#
+#     http://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 tensorflow as tf
+from zoo import init_nncontext
+from zoo.tfpark import TFOptimizer, TFDataset
+from bigdl.optim.optimizer import *
+import sys
+from tensorflow.keras.models import Model
+from tensorflow.keras.layers import *
+
+from bigdl.dataset import mnist
+from bigdl.dataset.transformer import *
+
+
+def main(max_epoch, data_num):
+    sc = init_nncontext()
+
+    # get data, pre-process and create TFDataset
+    def get_data_rdd(dataset):
+        (images_data, labels_data) = mnist.read_data_sets("/tmp/mnist", dataset)
+        image_rdd = sc.parallelize(images_data[:data_num])
+        labels_rdd = sc.parallelize(labels_data[:data_num])
+        rdd = image_rdd.zip(labels_rdd) \
+            .map(lambda rec_tuple: [normalizer(rec_tuple[0], mnist.TRAIN_MEAN, mnist.TRAIN_STD),
+                                    np.array(rec_tuple[1])])
+        return rdd
+
+    training_rdd = get_data_rdd("train")
+    testing_rdd = get_data_rdd("test")
+    dataset = TFDataset.from_rdd(training_rdd,
+                                 names=["features", "labels"],
+                                 shapes=[[28, 28, 1], []],
+                                 types=[tf.float32, tf.int32],
+                                 batch_size=280,
+                                 val_rdd=testing_rdd
+                                 )
+
+    data = Input(shape=[28, 28, 1])
+
+    x = Flatten()(data)
+    x = Dense(64, activation='relu')(x)
+    x = Dense(64, activation='relu')(x)
+    predictions = Dense(10, activation='softmax')(x)
+
+    model = Model(inputs=data, outputs=predictions)
+
+    model.compile(optimizer='rmsprop',
+                  loss='sparse_categorical_crossentropy',
+                  metrics=['accuracy'])
+
+    optimizer = TFOptimizer.from_keras(model, dataset)
+
+    optimizer.set_train_summary(TrainSummary("/tmp/mnist_log", "mnist"))
+    optimizer.set_val_summary(ValidationSummary("/tmp/mnist_log", "mnist"))
+    # kick off training
+    optimizer.optimize(end_trigger=MaxEpoch(max_epoch))
+
+    model.save_weights("/tmp/mnist_keras.h5")
+
+if __name__ == '__main__':
+
+    max_epoch = 5
+    data_num = 60000
+
+    if len(sys.argv) > 1:
+        max_epoch = int(sys.argv[1])
+        data_num = int(sys.argv[2])
+    main(max_epoch, data_num)