From ecc942f48070a445aab2264746cd36b3e5487966 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Wed, 25 Sep 2019 23:47:23 -0400
Subject: [PATCH 01/24] Cortex python client

---
 .../cortex/client/cortex/__init__.py          |   0
 pkg/workloads/cortex/client/cortex/client.py  | 109 ++++++++++++++++++
 pkg/workloads/cortex/client/setup.py          |  12 ++
 pkg/workloads/cortex/lib/context.py           |  22 +++-
 pkg/workloads/cortex/lib/requirements.txt     |   1 +
 5 files changed, 140 insertions(+), 4 deletions(-)
 create mode 100644 pkg/workloads/cortex/client/cortex/__init__.py
 create mode 100644 pkg/workloads/cortex/client/cortex/client.py
 create mode 100644 pkg/workloads/cortex/client/setup.py

diff --git a/pkg/workloads/cortex/client/cortex/__init__.py b/pkg/workloads/cortex/client/cortex/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
new file mode 100644
index 0000000000..624f48749b
--- /dev/null
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -0,0 +1,109 @@
+# Copyright 2019 Cortex Labs, Inc.
+#
+# 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 pathlib
+import os
+import types
+import subprocess
+import sys
+import shutil
+import yaml
+import urllib.parse
+
+import dill
+import requests
+
+
+class Client(object):
+    def __init__(
+        self,
+        aws_access_key_id,
+        aws_secret_access_key,
+        operator_url,
+        deployment_name,
+        workspace="./workspace",
+    ):
+        self.operator_url = operator_url
+        self.deployment_name = deployment_name
+        self.workspace = workspace
+        self.aws_access_key_id = aws_access_key_id
+        self.aws_secret_access_key = aws_secret_access_key
+        self.headers = {
+            "CortexAPIVersion": "master",
+            "Authorization": "CortexAWS {}|{}".format(
+                self.aws_access_key_id, self.aws_secret_access_key
+            ),
+        }
+
+        pathlib.Path(self.workspace).mkdir(parents=True, exist_ok=True)
+
+    def deploy(self, api_name, model_path, pre_inference, post_inference):
+        working_dir = os.path.join(self.workspace, self.deployment_name)
+        api_working_dir = os.path.join(working_dir, api_name)
+        pathlib.Path(api_working_dir).mkdir(parents=True, exist_ok=True)
+
+        reqs = subprocess.check_output([sys.executable, "-m", "pip", "freeze"])
+
+        with open(os.path.join(api_working_dir, "requirements.txt"), "w") as f:
+            f.writelines(reqs.decode())
+
+        with open(os.path.join(api_working_dir, "request_handler.pickle"), "wb") as f:
+            dill.dump(
+                {"pre_inference": pre_inference, "post_inference": post_inference}, f, recurse=True
+            )
+
+        cortex_config = [
+            {"kind": "deployment", "name": self.deployment_name},
+            {
+                "kind": "api",
+                "model": model_path,
+                "name": api_name,
+                "request_handler": "request_handler.pickle",
+            },
+        ]
+
+        cortex_yaml_path = os.path.join(working_dir, "cortex.yaml")
+        with open(cortex_yaml_path, "w") as f:
+            f.write(yaml.dump(cortex_config))
+
+        shutil.make_archive("project", "zip", api_working_dir)
+        project_zip_path = os.path.join(working_dir, "project.zip")
+
+        queries = {"force": "false", "ignoreCache": "false"}
+
+        with open(cortex_yaml_path, "rb") as config, open(project_zip_path, "rb") as project:
+            files = {"cortex.yaml": config, "project.zip": project}
+            print(urllib.parse.urljoin(self.operator_url, "deploy"))
+            resp = requests.post(
+                urllib.parse.urljoin(self.operator_url, "deploy"),
+                params=queries,
+                files=files,
+                headers=self.headers,
+                verify=False,
+            )
+            print(resp.json())
+
+
+    def get(self, api_name):
+        queries = {"appName": self.deployment_name}
+
+        resp = requests.get(
+            urllib.parse.urljoin(self.operator_url, "get"),
+            params=queries,
+            headers=self.headers,
+            verify=False,
+        )
+
+        resources = resp.json()
+        urllib.parse.urljoin(resources['apis_base_url'], resources[])
diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
new file mode 100644
index 0000000000..2fe9ac83e7
--- /dev/null
+++ b/pkg/workloads/cortex/client/setup.py
@@ -0,0 +1,12 @@
+from setuptools import setup
+
+
+setup(
+    name="cortex",
+    version="0.8.0",
+    description="",
+    author="Cortex Labs",
+    author_email="dev@cortexlabs.com",
+    install_requires=["dill==0.3.0", "requests==2.21.0"],
+    setup_requires=["setuptools"],
+)
diff --git a/pkg/workloads/cortex/lib/context.py b/pkg/workloads/cortex/lib/context.py
index 1106b58deb..0b6efe45e1 100644
--- a/pkg/workloads/cortex/lib/context.py
+++ b/pkg/workloads/cortex/lib/context.py
@@ -15,8 +15,10 @@
 import os
 import imp
 import inspect
+
 import boto3
 import datadog
+import dill
 
 from cortex import consts
 from cortex.lib import util
@@ -108,10 +110,22 @@ def download_python_file(self, impl_key, module_name):
 
     def load_module(self, module_prefix, module_name, impl_path):
         full_module_name = "{}_{}".format(module_prefix, module_name)
-        try:
-            impl = imp.load_source(full_module_name, impl_path)
-        except Exception as e:
-            raise UserException("unable to load python file", str(e)) from e
+
+        if impl_path.endswith("pickle"):
+            impl = imp.new_module(full_module_name)
+
+            logger.info(impl_path)
+            with open(impl_path, "rb") as pickle_file:
+                pickled_dict = dill.load(pickle_file)
+                logger.info(pickled_dict)
+                for key in pickled_dict:
+                    logger.info(key)
+                    setattr(impl, key, pickled_dict[key])
+        else:
+            try:
+                impl = imp.load_source(full_module_name, impl_path)
+            except Exception as e:
+                raise UserException("unable to load python file", str(e)) from e
 
         return impl
 
diff --git a/pkg/workloads/cortex/lib/requirements.txt b/pkg/workloads/cortex/lib/requirements.txt
index 3a721783a3..af7fa9db73 100644
--- a/pkg/workloads/cortex/lib/requirements.txt
+++ b/pkg/workloads/cortex/lib/requirements.txt
@@ -4,3 +4,4 @@ numpy==1.17.2
 json_tricks==3.13.2
 requests==2.22.0
 datadog==0.30.0
+dill==0.3.0
\ No newline at end of file

From 5d383f818485159239d95a9ebf037ab57e7360ae Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 11:38:12 -0400
Subject: [PATCH 02/24] Progress

---
 pkg/workloads/cortex/client/cortex/client.py | 34 ++++++++------------
 1 file changed, 13 insertions(+), 21 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 624f48749b..2f37476fba 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -1,17 +1,3 @@
-# Copyright 2019 Cortex Labs, Inc.
-#
-# 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 pathlib
 import os
 import types
@@ -77,8 +63,10 @@ def deploy(self, api_name, model_path, pre_inference, post_inference):
         with open(cortex_yaml_path, "w") as f:
             f.write(yaml.dump(cortex_config))
 
-        shutil.make_archive("project", "zip", api_working_dir)
-        project_zip_path = os.path.join(working_dir, "project.zip")
+        print(api_working_dir)
+        project_zip_path = os.path.join(working_dir, "project")
+        shutil.make_archive(project_zip_path, "zip", api_working_dir)
+        project_zip_path += ".zip"
 
         queries = {"force": "false", "ignoreCache": "false"}
 
@@ -94,16 +82,20 @@ def deploy(self, api_name, model_path, pre_inference, post_inference):
             )
             print(resp.json())
 
-
-    def get(self, api_name):
+    def get_endpoint(self, api_name):
         queries = {"appName": self.deployment_name}
 
         resp = requests.get(
-            urllib.parse.urljoin(self.operator_url, "get"),
+            urllib.parse.urljoin(self.operator_url, "resources"),
             params=queries,
             headers=self.headers,
             verify=False,
         )
-
+        print(resp)
         resources = resp.json()
-        urllib.parse.urljoin(resources['apis_base_url'], resources[])
+        print(
+            urllib.parse.urljoin(
+                resources["apis_base_url"],
+                resources["api_name_statuses"][api_name]["active_status"]["path"],
+            )
+        )

From 8cfe00293e81566f14fa157d6ffc7f4b59308729 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 11:48:02 -0400
Subject: [PATCH 03/24] Enforce python version in setup.py

---
 pkg/workloads/cortex/client/setup.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index 2fe9ac83e7..2c093f8916 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -9,4 +9,5 @@
     author_email="dev@cortexlabs.com",
     install_requires=["dill==0.3.0", "requests==2.21.0"],
     setup_requires=["setuptools"],
+    python_requires=">=3.6",
 )

From afb98347a0298bc1776430547f5f920f3f532939 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 11:56:47 -0400
Subject: [PATCH 04/24] another version

---
 pkg/workloads/cortex/client/setup.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index 2c093f8916..19b4ffbb91 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -9,5 +9,5 @@
     author_email="dev@cortexlabs.com",
     install_requires=["dill==0.3.0", "requests==2.21.0"],
     setup_requires=["setuptools"],
-    python_requires=">=3.6",
+    python_requires=[">=3.6", "==2.7.*"],
 )

From 769512dda10b68e2ecf7728692367fdbfa3f2981 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 12:00:52 -0400
Subject: [PATCH 05/24] Again

---
 pkg/workloads/cortex/client/setup.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index 19b4ffbb91..d3cb19ecc6 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -9,5 +9,5 @@
     author_email="dev@cortexlabs.com",
     install_requires=["dill==0.3.0", "requests==2.21.0"],
     setup_requires=["setuptools"],
-    python_requires=[">=3.6", "==2.7.*"],
+    python_requires=[">=3.6", "==2.7"],
 )

From 23334cda253c34fa121b32b55ac50dbf546e7bb5 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 12:03:45 -0400
Subject: [PATCH 06/24] Back to 3.6

---
 pkg/workloads/cortex/client/setup.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index d3cb19ecc6..2c093f8916 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -9,5 +9,5 @@
     author_email="dev@cortexlabs.com",
     install_requires=["dill==0.3.0", "requests==2.21.0"],
     setup_requires=["setuptools"],
-    python_requires=[">=3.6", "==2.7"],
+    python_requires=">=3.6",
 )

From 5d6cb68cbe2b7a49e77bdaf67d5223bdd3b519d3 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 12:27:10 -0400
Subject: [PATCH 07/24] Remove python version

---
 pkg/workloads/cortex/client/setup.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index 2c093f8916..5fe0940e1c 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -1,4 +1,4 @@
-from setuptools import setup
+from setuptools import setup, find_packages
 
 
 setup(
@@ -9,5 +9,5 @@
     author_email="dev@cortexlabs.com",
     install_requires=["dill==0.3.0", "requests==2.21.0"],
     setup_requires=["setuptools"],
-    python_requires=">=3.6",
+    packages=find_packages(),
 )

From 75382253b74e5086336eec7423c3e1d88d8dcfaf Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 15:42:14 -0400
Subject: [PATCH 08/24] Get endpoint in deploy

---
 .../cortex/client/cortex/__init__.py          | 15 ++++++
 pkg/workloads/cortex/client/cortex/client.py  | 51 ++++++++++---------
 2 files changed, 43 insertions(+), 23 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/__init__.py b/pkg/workloads/cortex/client/cortex/__init__.py
index e69de29bb2..4005eec0fa 100644
--- a/pkg/workloads/cortex/client/cortex/__init__.py
+++ b/pkg/workloads/cortex/client/cortex/__init__.py
@@ -0,0 +1,15 @@
+# Copyright 2019 Cortex Labs, Inc.
+#
+# 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.
+
+from cortex.client import Client
diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 2f37476fba..ab56c121f3 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -13,15 +13,9 @@
 
 class Client(object):
     def __init__(
-        self,
-        aws_access_key_id,
-        aws_secret_access_key,
-        operator_url,
-        deployment_name,
-        workspace="./workspace",
+        self, aws_access_key_id, aws_secret_access_key, operator_url, workspace="./workspace"
     ):
         self.operator_url = operator_url
-        self.deployment_name = deployment_name
         self.workspace = workspace
         self.aws_access_key_id = aws_access_key_id
         self.aws_secret_access_key = aws_secret_access_key
@@ -34,8 +28,8 @@ def __init__(
 
         pathlib.Path(self.workspace).mkdir(parents=True, exist_ok=True)
 
-    def deploy(self, api_name, model_path, pre_inference, post_inference):
-        working_dir = os.path.join(self.workspace, self.deployment_name)
+    def deploy(self, deployment_name, api_name, model_path, pre_inference, post_inference):
+        working_dir = os.path.join(self.workspace, deployment_name)
         api_working_dir = os.path.join(working_dir, api_name)
         pathlib.Path(api_working_dir).mkdir(parents=True, exist_ok=True)
 
@@ -50,7 +44,7 @@ def deploy(self, api_name, model_path, pre_inference, post_inference):
             )
 
         cortex_config = [
-            {"kind": "deployment", "name": self.deployment_name},
+            {"kind": "deployment", "name": deployment_name},
             {
                 "kind": "api",
                 "model": model_path,
@@ -63,16 +57,16 @@ def deploy(self, api_name, model_path, pre_inference, post_inference):
         with open(cortex_yaml_path, "w") as f:
             f.write(yaml.dump(cortex_config))
 
-        print(api_working_dir)
         project_zip_path = os.path.join(working_dir, "project")
         shutil.make_archive(project_zip_path, "zip", api_working_dir)
         project_zip_path += ".zip"
 
         queries = {"force": "false", "ignoreCache": "false"}
 
+        deployment_status = {}
+
         with open(cortex_yaml_path, "rb") as config, open(project_zip_path, "rb") as project:
             files = {"cortex.yaml": config, "project.zip": project}
-            print(urllib.parse.urljoin(self.operator_url, "deploy"))
             resp = requests.post(
                 urllib.parse.urljoin(self.operator_url, "deploy"),
                 params=queries,
@@ -80,10 +74,28 @@ def deploy(self, api_name, model_path, pre_inference, post_inference):
                 headers=self.headers,
                 verify=False,
             )
-            print(resp.json())
+            deployment_status = resp.json()
+
+            if resp.status_code / 100 != 2:
+                return deployment_status
+
+            if resp.status_code / 100 == 2:
+                print(deployment_status["message"])
+                endpoint_response = self.get_endpoint(deployment_name, api_name)
+
+                if endpoint_response.status_code / 100 != 2:
+                    return endpoint_response
 
-    def get_endpoint(self, api_name):
-        queries = {"appName": self.deployment_name}
+                resources = endpoint_response.json()
+                return {
+                    "url": urllib.parse.urljoin(
+                        resources["apis_base_url"],
+                        resources["api_name_statuses"][api_name]["active_status"]["path"],
+                    )
+                }
+
+    def get_endpoint(self, deployment_name, api_name):
+        queries = {"appName": deployment_name}
 
         resp = requests.get(
             urllib.parse.urljoin(self.operator_url, "resources"),
@@ -91,11 +103,4 @@ def get_endpoint(self, api_name):
             headers=self.headers,
             verify=False,
         )
-        print(resp)
-        resources = resp.json()
-        print(
-            urllib.parse.urljoin(
-                resources["apis_base_url"],
-                resources["api_name_statuses"][api_name]["active_status"]["path"],
-            )
-        )
+        return resp

From 6369e2ceb0f5113f2366da7aa5242e37b79fdc8b Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 15:48:47 -0400
Subject: [PATCH 09/24] Fix linting

---
 pkg/workloads/cortex/client/cortex/client.py | 14 ++++++++++++++
 pkg/workloads/cortex/client/setup.py         | 15 ++++++++++++++-
 pkg/workloads/cortex/lib/requirements.txt    |  2 +-
 3 files changed, 29 insertions(+), 2 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index ab56c121f3..9bb03699ce 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -1,3 +1,17 @@
+# Copyright 2019 Cortex Labs, Inc.
+#
+# 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 pathlib
 import os
 import types
diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index 5fe0940e1c..fc771d8fb8 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -1,5 +1,18 @@
-from setuptools import setup, find_packages
+# Copyright 2019 Cortex Labs, Inc.
+#
+# 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.
 
+from setuptools import setup, find_packages
 
 setup(
     name="cortex",
diff --git a/pkg/workloads/cortex/lib/requirements.txt b/pkg/workloads/cortex/lib/requirements.txt
index af7fa9db73..fda1ae01b3 100644
--- a/pkg/workloads/cortex/lib/requirements.txt
+++ b/pkg/workloads/cortex/lib/requirements.txt
@@ -4,4 +4,4 @@ numpy==1.17.2
 json_tricks==3.13.2
 requests==2.22.0
 datadog==0.30.0
-dill==0.3.0
\ No newline at end of file
+dill==0.3.0

From f65d6b64169da18188201262e0e77990a2821fe7 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 16:24:49 -0400
Subject: [PATCH 10/24] Raise exception when requests fail

---
 pkg/workloads/cortex/client/cortex/client.py | 67 +++++++++++---------
 1 file changed, 37 insertions(+), 30 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 9bb03699ce..236f04c149 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -23,6 +23,7 @@
 
 import dill
 import requests
+from requests.exceptions import HTTPError
 
 
 class Client(object):
@@ -81,40 +82,46 @@ def deploy(self, deployment_name, api_name, model_path, pre_inference, post_infe
 
         with open(cortex_yaml_path, "rb") as config, open(project_zip_path, "rb") as project:
             files = {"cortex.yaml": config, "project.zip": project}
-            resp = requests.post(
-                urllib.parse.urljoin(self.operator_url, "deploy"),
-                params=queries,
-                files=files,
-                headers=self.headers,
-                verify=False,
-            )
-            deployment_status = resp.json()
-
-            if resp.status_code / 100 != 2:
-                return deployment_status
+            try:
+                resp = requests.post(
+                    urllib.parse.urljoin(self.operator_url, "deploy"),
+                    params=queries,
+                    files=files,
+                    headers=self.headers,
+                    verify=False,
+                )
+                resp.raise_for_status()
+                deployment_status = resp.json()
+            except HTTPError as err:
+                resp = err.response
+                if "error" in resp.json():
+                    raise Exception(resp.json()["error"]) from err
+                raise
 
             if resp.status_code / 100 == 2:
-                print(deployment_status["message"])
-                endpoint_response = self.get_endpoint(deployment_name, api_name)
-
-                if endpoint_response.status_code / 100 != 2:
-                    return endpoint_response
+                return self.get_endpoint(deployment_name, api_name)
 
-                resources = endpoint_response.json()
-                return {
-                    "url": urllib.parse.urljoin(
-                        resources["apis_base_url"],
-                        resources["api_name_statuses"][api_name]["active_status"]["path"],
-                    )
-                }
+            return None
 
     def get_endpoint(self, deployment_name, api_name):
         queries = {"appName": deployment_name}
 
-        resp = requests.get(
-            urllib.parse.urljoin(self.operator_url, "resources"),
-            params=queries,
-            headers=self.headers,
-            verify=False,
-        )
-        return resp
+        try:
+            resp = requests.get(
+                urllib.parse.urljoin(self.operator_url, "resources"),
+                params=queries,
+                headers=self.headers,
+                verify=False,
+            )
+            resp.raise_for_status()
+
+            resources = resp.json()
+            return urllib.parse.urljoin(
+                resources["apis_base_url"],
+                resources["api_name_statuses"][api_name]["active_status"]["path"],
+            )
+        except HTTPError as err:
+            resp = err.response
+            if "error" in resp.json():
+                raise Exception(resp.json()["error"]) from err
+            raise

From 64dba0b880079b43e44b20ea0d582be751446feb Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 16:47:00 -0400
Subject: [PATCH 11/24] Remove logging statements

---
 pkg/workloads/cortex/client/cortex/client.py | 2 +-
 pkg/workloads/cortex/lib/context.py          | 3 ---
 2 files changed, 1 insertion(+), 4 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 236f04c149..8a14925843 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -98,7 +98,7 @@ def deploy(self, deployment_name, api_name, model_path, pre_inference, post_infe
                     raise Exception(resp.json()["error"]) from err
                 raise
 
-            if resp.status_code / 100 == 2:
+            if resp.status_code == 200:
                 return self.get_endpoint(deployment_name, api_name)
 
             return None
diff --git a/pkg/workloads/cortex/lib/context.py b/pkg/workloads/cortex/lib/context.py
index 0b6efe45e1..5c3dedba96 100644
--- a/pkg/workloads/cortex/lib/context.py
+++ b/pkg/workloads/cortex/lib/context.py
@@ -114,12 +114,9 @@ def load_module(self, module_prefix, module_name, impl_path):
         if impl_path.endswith("pickle"):
             impl = imp.new_module(full_module_name)
 
-            logger.info(impl_path)
             with open(impl_path, "rb") as pickle_file:
                 pickled_dict = dill.load(pickle_file)
-                logger.info(pickled_dict)
                 for key in pickled_dict:
-                    logger.info(key)
                     setattr(impl, key, pickled_dict[key])
         else:
             try:

From a8f5cd0c0518ce72a8a657b9c30e874f42c5a511 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 16:58:04 -0400
Subject: [PATCH 12/24] Cleanup rount 1

---
 pkg/workloads/cortex/client/cortex/client.py | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 8a14925843..6ad8967d48 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -78,8 +78,6 @@ def deploy(self, deployment_name, api_name, model_path, pre_inference, post_infe
 
         queries = {"force": "false", "ignoreCache": "false"}
 
-        deployment_status = {}
-
         with open(cortex_yaml_path, "rb") as config, open(project_zip_path, "rb") as project:
             files = {"cortex.yaml": config, "project.zip": project}
             try:
@@ -91,7 +89,6 @@ def deploy(self, deployment_name, api_name, model_path, pre_inference, post_infe
                     verify=False,
                 )
                 resp.raise_for_status()
-                deployment_status = resp.json()
             except HTTPError as err:
                 resp = err.response
                 if "error" in resp.json():

From eb4a6f620e3b6605e54581864b9165ca6ae6ea1d Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 18:07:59 -0400
Subject: [PATCH 13/24] Print resources

---
 pkg/workloads/cortex/client/cortex/client.py | 2 ++
 pkg/workloads/cortex/client/setup.py         | 2 +-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 6ad8967d48..0829e79cf9 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -24,6 +24,7 @@
 import dill
 import requests
 from requests.exceptions import HTTPError
+import msgpack
 
 
 class Client(object):
@@ -113,6 +114,7 @@ def get_endpoint(self, deployment_name, api_name):
             resp.raise_for_status()
 
             resources = resp.json()
+            print(resources)
             return urllib.parse.urljoin(
                 resources["apis_base_url"],
                 resources["api_name_statuses"][api_name]["active_status"]["path"],
diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index fc771d8fb8..601ebb6fd5 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -20,7 +20,7 @@
     description="",
     author="Cortex Labs",
     author_email="dev@cortexlabs.com",
-    install_requires=["dill==0.3.0", "requests==2.21.0"],
+    install_requires=["dill==0.3.0", "requests==2.21.0", "msgpack==0.6.1"],
     setup_requires=["setuptools"],
     packages=find_packages(),
 )

From 94e56ef813df89fb64ad77b271f0e17d98a34934 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Thu, 26 Sep 2019 19:34:24 -0400
Subject: [PATCH 14/24] Get path from context

---
 pkg/workloads/cortex/client/cortex/client.py | 48 ++++++++++----------
 1 file changed, 23 insertions(+), 25 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 0829e79cf9..415709b662 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -20,6 +20,7 @@
 import shutil
 import yaml
 import urllib.parse
+import base64
 
 import dill
 import requests
@@ -97,30 +98,27 @@ def deploy(self, deployment_name, api_name, model_path, pre_inference, post_infe
                 raise
 
             if resp.status_code == 200:
-                return self.get_endpoint(deployment_name, api_name)
+                queries = {"appName": deployment_name}
+
+                try:
+                    resp = requests.get(
+                        urllib.parse.urljoin(self.operator_url, "resources"),
+                        params=queries,
+                        headers=self.headers,
+                        verify=False,
+                    )
+                    resp.raise_for_status()
+                    resources = resp.json()
+                    b64_encoded_context = resources["context"]
+                    context_msgpack_bytestring = base64.b64decode(b64_encoded_context)
+                    ctx = msgpack.loads(context_msgpack_bytestring, raw=False)
+                    return urllib.parse.urljoin(
+                        resources["apis_base_url"], ctx["apis"][api_name]["path"]
+                    )
+                except HTTPError as err:
+                    resp = err.response
+                    if "error" in resp.json():
+                        raise Exception(resp.json()["error"]) from err
+                    raise
 
             return None
-
-    def get_endpoint(self, deployment_name, api_name):
-        queries = {"appName": deployment_name}
-
-        try:
-            resp = requests.get(
-                urllib.parse.urljoin(self.operator_url, "resources"),
-                params=queries,
-                headers=self.headers,
-                verify=False,
-            )
-            resp.raise_for_status()
-
-            resources = resp.json()
-            print(resources)
-            return urllib.parse.urljoin(
-                resources["apis_base_url"],
-                resources["api_name_statuses"][api_name]["active_status"]["path"],
-            )
-        except HTTPError as err:
-            resp = err.response
-            if "error" in resp.json():
-                raise Exception(resp.json()["error"]) from err
-            raise

From 38812bb1aa9d4cb273b779d2e7fdb049a7bc2ac2 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Fri, 27 Sep 2019 16:56:16 -0400
Subject: [PATCH 15/24] Add docs and respond PR comments

---
 dev/versions.md                              |   1 +
 docs/cluster/python-client.md                |  44 ++++++++
 pkg/operator/api/schema/schema.go            |   4 +-
 pkg/operator/endpoints/deploy.go             |  23 ++--
 pkg/workloads/cortex/client/cortex/client.py | 112 +++++++++++--------
 pkg/workloads/cortex/client/setup.py         |   2 +-
 pkg/workloads/cortex/lib/context.py          |  15 ++-
 7 files changed, 140 insertions(+), 61 deletions(-)
 create mode 100644 docs/cluster/python-client.md

diff --git a/dev/versions.md b/dev/versions.md
index c04bb4af42..57df5fff6b 100644
--- a/dev/versions.md
+++ b/dev/versions.md
@@ -103,6 +103,7 @@ Note: it's ok if example training notebooks aren't upgraded, as long as the expo
    * [flask](https://pypi.org/project/flask/)
    * [flask-api](https://pypi.org/project/flask-api/)
    * [waitress](https://pypi.org/project/waitress/)
+   * [dill](https://pypi.org/project/dill/)
 1. Update the versions listed in "Pre-installed Packages" in `request-handlers.py`
 
 ## Istio
diff --git a/docs/cluster/python-client.md b/docs/cluster/python-client.md
new file mode 100644
index 0000000000..5999d08c0a
--- /dev/null
+++ b/docs/cluster/python-client.md
@@ -0,0 +1,44 @@
+# Python Client
+
+Python Client can be used to programmatically deploy to a Cortex Cluster.
+
+<!-- CORTEX_VERSION_MINOR -->
+```
+pip install git+https://github.com/cortexlabs/cortex.git@master#egg=cortex\&subdirectory=pkg/workloads/cortex/client
+```
+
+Python client needs to be initialized with AWS credentials and a url to the operator your Cortex cluster.
+
+```python
+from cortex import Client
+
+cortex = Client(
+    aws_access_key_id="<string>",  # AWS access key associated with the account that created the Cortex cluster
+    aws_secret_access_key="<string>",  # AWS secrey key associated with the provided AWS access key  
+    operator_url="<string>" # Operator URL of your cortex cluster
+)
+
+api_url = cortex.deploy(
+    deployment_name="<string>",  # Deployment name (required)
+    api_name="<string>",  # API name (required)
+    model_path="<string>",  # S3 path to an exported model (required)
+    pre_inference=callable,  # a function used to prepare requests for model input
+    post_inference=callable,  # a function used to prepare model output for response
+    model_format="<string>",  # model format, must be "tensorflow" or "onnx" (default: "onnx" if model path ends with .onnx, "tensorflow" if model path ends with .zip or is a directory)
+    tf_serving_key="<string>"  # name of the signature def to use for prediction (required if your model has more than one signature def)
+)
+```
+
+`api_url` is the url to the deployed API, it accepts JSON POST requests.
+
+```python
+import requests
+
+sample = {
+    "feature_1" [0,1,2]
+}
+
+resp = requests.post(api_url, json=sample)
+
+resp.json() # your model prediction
+```
\ No newline at end of file
diff --git a/pkg/operator/api/schema/schema.go b/pkg/operator/api/schema/schema.go
index 1ded820ae0..05735d3291 100644
--- a/pkg/operator/api/schema/schema.go
+++ b/pkg/operator/api/schema/schema.go
@@ -24,7 +24,9 @@ import (
 )
 
 type DeployResponse struct {
-	Message string `json:"message"`
+	Message     string           `json:"message"`
+	Context     *context.Context `json:"context"`
+	APIsBaseURL string           `json:"apis_base_url"`
 }
 
 type DeleteResponse struct {
diff --git a/pkg/operator/endpoints/deploy.go b/pkg/operator/endpoints/deploy.go
index 968563c7ec..5a39bdaf9a 100644
--- a/pkg/operator/endpoints/deploy.go
+++ b/pkg/operator/endpoints/deploy.go
@@ -117,20 +117,29 @@ func Deploy(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	apisBaseURL, err := workloads.APIsBaseURL()
+	if err != nil {
+		RespondError(w, err)
+		return
+	}
+
+	deployResponse := schema.DeployResponse{Context: ctx, APIsBaseURL: apisBaseURL}
 	switch {
 	case isUpdating && ignoreCache:
-		Respond(w, schema.DeployResponse{Message: ResCachedDeletedDeploymentStarted})
+		deployResponse.Message = ResCachedDeletedDeploymentStarted
 	case isUpdating && !ignoreCache:
-		Respond(w, schema.DeployResponse{Message: ResDeploymentUpdated})
+		deployResponse.Message = ResDeploymentUpdated
 	case !isUpdating && ignoreCache:
-		Respond(w, schema.DeployResponse{Message: ResCachedDeletedDeploymentStarted})
+		deployResponse.Message = ResCachedDeletedDeploymentStarted
 	case !isUpdating && !ignoreCache && existingCtx == nil:
-		Respond(w, schema.DeployResponse{Message: ResDeploymentStarted})
+		deployResponse.Message = ResDeploymentStarted
 	case !isUpdating && !ignoreCache && existingCtx != nil && !fullCtxMatch:
-		Respond(w, schema.DeployResponse{Message: ResDeploymentUpdated})
+		deployResponse.Message = ResDeploymentUpdated
 	case !isUpdating && !ignoreCache && existingCtx != nil && fullCtxMatch:
-		Respond(w, schema.DeployResponse{Message: ResDeploymentUpToDate})
+		deployResponse.Message = ResDeploymentUpToDate
 	default:
-		Respond(w, schema.DeployResponse{Message: ResDeploymentUpdated})
+		deployResponse.Message = ResDeploymentUpdated
 	}
+
+	Respond(w, deployResponse)
 }
diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 415709b662..9757b42899 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -13,6 +13,7 @@
 # limitations under the License.
 
 import pathlib
+from pathlib import Path
 import os
 import types
 import subprocess
@@ -29,11 +30,17 @@
 
 
 class Client(object):
-    def __init__(
-        self, aws_access_key_id, aws_secret_access_key, operator_url, workspace="./workspace"
-    ):
+    def __init__(self, aws_access_key_id, aws_secret_access_key, operator_url):
+        """Initialize a Client to a Cortex Operator
+        
+        Args:
+            aws_access_key_id (string): AWS access key associated with the account that created the Cortex Cluster
+            aws_secret_access_key (string): AWS secrey key associated with the provided AWS access key  
+            operator_url (string): Operator URL of your cortex cluster
+        """
+
         self.operator_url = operator_url
-        self.workspace = workspace
+        self.workspace = str(Path.home() / ".cortex" / "workspace")
         self.aws_access_key_id = aws_access_key_id
         self.aws_secret_access_key = aws_secret_access_key
         self.headers = {
@@ -45,30 +52,63 @@ def __init__(
 
         pathlib.Path(self.workspace).mkdir(parents=True, exist_ok=True)
 
-    def deploy(self, deployment_name, api_name, model_path, pre_inference, post_inference):
+    def deploy(
+        self,
+        deployment_name,
+        api_name,
+        model_path,
+        pre_inference=None,
+        post_inference=None,
+        model_format=None,
+        tf_serving_key=None,
+    ):
+        """Deploy an API
+        
+        Args:
+            deployment_name (string): Deployment name
+            api_name (string): API name
+            model_path (string): S3 path to an exported model
+            pre_inference (function, optional): a function used to prepare requests for model input. Defaults to None.
+            post_inference (function, optional): a function used to prepare model output for response. Defaults to None.
+            model_format (string, optional): model format, must be "tensorflow" or "onnx" (default: "onnx" if model path ends with .onnx, "tensorflow" if model path ends with .zip or is a directory)
+            tf_serving_key (string, optional): name of the signature def to use for prediction (required if your model has more than one signature def)
+        
+        Returns:
+            string: url to the deployed API
+        """
+
         working_dir = os.path.join(self.workspace, deployment_name)
         api_working_dir = os.path.join(working_dir, api_name)
         pathlib.Path(api_working_dir).mkdir(parents=True, exist_ok=True)
 
-        reqs = subprocess.check_output([sys.executable, "-m", "pip", "freeze"])
+        api_config = {
+            "kind": "api",
+            "model": model_path,
+            "name": api_name,
+            "model_format": model_format,
+            "tf_serving_key": tf_serving_key,
+        }
+
+        if pre_inference is not None or post_inference is not None:
+            reqs = subprocess.check_output([sys.executable, "-m", "pip", "freeze"])
+
+            with open(os.path.join(api_working_dir, "requirements.txt"), "w") as f:
+                f.writelines(reqs.decode())
+
+            handlers = {}
+
+            if pre_inference is not None:
+                handlers["pre_inference"] = pre_inference
+
+            if post_inference is not None:
+                handlers["post_inference"] = post_inference
 
-        with open(os.path.join(api_working_dir, "requirements.txt"), "w") as f:
-            f.writelines(reqs.decode())
+            with open(os.path.join(api_working_dir, "request_handler.pickle"), "wb") as f:
+                dill.dump(handlers, f, recurse=True)
 
-        with open(os.path.join(api_working_dir, "request_handler.pickle"), "wb") as f:
-            dill.dump(
-                {"pre_inference": pre_inference, "post_inference": post_inference}, f, recurse=True
-            )
+            api_config["request_handler"] = "request_handler.pickle"
 
-        cortex_config = [
-            {"kind": "deployment", "name": deployment_name},
-            {
-                "kind": "api",
-                "model": model_path,
-                "name": api_name,
-                "request_handler": "request_handler.pickle",
-            },
-        ]
+        cortex_config = [{"kind": "deployment", "name": deployment_name}, api_config]
 
         cortex_yaml_path = os.path.join(working_dir, "cortex.yaml")
         with open(cortex_yaml_path, "w") as f:
@@ -91,34 +131,14 @@ def deploy(self, deployment_name, api_name, model_path, pre_inference, post_infe
                     verify=False,
                 )
                 resp.raise_for_status()
+                resources = resp.json()
             except HTTPError as err:
                 resp = err.response
                 if "error" in resp.json():
                     raise Exception(resp.json()["error"]) from err
                 raise
 
-            if resp.status_code == 200:
-                queries = {"appName": deployment_name}
-
-                try:
-                    resp = requests.get(
-                        urllib.parse.urljoin(self.operator_url, "resources"),
-                        params=queries,
-                        headers=self.headers,
-                        verify=False,
-                    )
-                    resp.raise_for_status()
-                    resources = resp.json()
-                    b64_encoded_context = resources["context"]
-                    context_msgpack_bytestring = base64.b64decode(b64_encoded_context)
-                    ctx = msgpack.loads(context_msgpack_bytestring, raw=False)
-                    return urllib.parse.urljoin(
-                        resources["apis_base_url"], ctx["apis"][api_name]["path"]
-                    )
-                except HTTPError as err:
-                    resp = err.response
-                    if "error" in resp.json():
-                        raise Exception(resp.json()["error"]) from err
-                    raise
-
-            return None
+            b64_encoded_context = resources["context"]
+            context_msgpack_bytestring = base64.b64decode(b64_encoded_context)
+            ctx = msgpack.loads(context_msgpack_bytestring, raw=False)
+            return urllib.parse.urljoin(resources["apis_base_url"], ctx["apis"][api_name]["path"])
diff --git a/pkg/workloads/cortex/client/setup.py b/pkg/workloads/cortex/client/setup.py
index 601ebb6fd5..bd00510ca8 100644
--- a/pkg/workloads/cortex/client/setup.py
+++ b/pkg/workloads/cortex/client/setup.py
@@ -20,7 +20,7 @@
     description="",
     author="Cortex Labs",
     author_email="dev@cortexlabs.com",
-    install_requires=["dill==0.3.0", "requests==2.21.0", "msgpack==0.6.1"],
+    install_requires=["dill>=0.3.0", "requests>=2.20.0", "msgpack>=0.6.0"],
     setup_requires=["setuptools"],
     packages=find_packages(),
 )
diff --git a/pkg/workloads/cortex/lib/context.py b/pkg/workloads/cortex/lib/context.py
index 5c3dedba96..e10617fdd9 100644
--- a/pkg/workloads/cortex/lib/context.py
+++ b/pkg/workloads/cortex/lib/context.py
@@ -111,13 +111,16 @@ def download_python_file(self, impl_key, module_name):
     def load_module(self, module_prefix, module_name, impl_path):
         full_module_name = "{}_{}".format(module_prefix, module_name)
 
-        if impl_path.endswith("pickle"):
-            impl = imp.new_module(full_module_name)
+        if impl_path.endswith(".pickle"):
+            try:
+                impl = imp.new_module(full_module_name)
 
-            with open(impl_path, "rb") as pickle_file:
-                pickled_dict = dill.load(pickle_file)
-                for key in pickled_dict:
-                    setattr(impl, key, pickled_dict[key])
+                with open(impl_path, "rb") as pickle_file:
+                    pickled_dict = dill.load(pickle_file)
+                    for key in pickled_dict:
+                        setattr(impl, key, pickled_dict[key])
+            except Exception as e:
+                raise UserException("unable to load pickle", str(e)) from e
         else:
             try:
                 impl = imp.load_source(full_module_name, impl_path)

From e6745c65ffc8a43f4b17e7f464d5fb07e9270b06 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Fri, 27 Sep 2019 17:41:11 -0400
Subject: [PATCH 16/24] Use pathlib

---
 pkg/workloads/cortex/lib/util.py | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

diff --git a/pkg/workloads/cortex/lib/util.py b/pkg/workloads/cortex/lib/util.py
index e28685a1f0..3f4f6244da 100644
--- a/pkg/workloads/cortex/lib/util.py
+++ b/pkg/workloads/cortex/lib/util.py
@@ -23,6 +23,7 @@
 import zipfile
 import hashlib
 import msgpack
+import pathlib
 from copy import deepcopy
 from datetime import datetime
 
@@ -62,13 +63,7 @@ def snake_to_camel(input, sep="_", lower=True):
 
 
 def mkdir_p(dir_path):
-    try:
-        os.makedirs(dir_path)
-    except OSError as e:
-        if e.errno == errno.EEXIST and os.path.isdir(dir_path):
-            pass
-        else:
-            raise
+    pathlib.Path(dir_path).mkdir(parents=True, exist_ok=True)
 
 
 def rm_dir(dir_path):

From bb6d8d2b3978aec9c5466ba9d2aaacc641363788 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Fri, 27 Sep 2019 17:42:17 -0400
Subject: [PATCH 17/24] Drop articles

---
 docs/cluster/python-client.md | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/docs/cluster/python-client.md b/docs/cluster/python-client.md
index 5999d08c0a..b036995e47 100644
--- a/docs/cluster/python-client.md
+++ b/docs/cluster/python-client.md
@@ -19,11 +19,11 @@ cortex = Client(
 )
 
 api_url = cortex.deploy(
-    deployment_name="<string>",  # Deployment name (required)
+    deployment_name="<string>",  # deployment name (required)
     api_name="<string>",  # API name (required)
     model_path="<string>",  # S3 path to an exported model (required)
-    pre_inference=callable,  # a function used to prepare requests for model input
-    post_inference=callable,  # a function used to prepare model output for response
+    pre_inference=callable,  # function used to prepare requests for model input
+    post_inference=callable,  # function used to prepare model output for response
     model_format="<string>",  # model format, must be "tensorflow" or "onnx" (default: "onnx" if model path ends with .onnx, "tensorflow" if model path ends with .zip or is a directory)
     tf_serving_key="<string>"  # name of the signature def to use for prediction (required if your model has more than one signature def)
 )

From 421bd086259a68d5641c3ac64a0b61f251f7e520 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Fri, 27 Sep 2019 17:42:41 -0400
Subject: [PATCH 18/24] Drop leading articles in python docs

---
 pkg/workloads/cortex/client/cortex/client.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 9757b42899..ad5d5ca12a 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -65,11 +65,11 @@ def deploy(
         """Deploy an API
         
         Args:
-            deployment_name (string): Deployment name
+            deployment_name (string): deployment name
             api_name (string): API name
             model_path (string): S3 path to an exported model
-            pre_inference (function, optional): a function used to prepare requests for model input. Defaults to None.
-            post_inference (function, optional): a function used to prepare model output for response. Defaults to None.
+            pre_inference (function, optional): function used to prepare requests for model input. Defaults to None.
+            post_inference (function, optional): function used to prepare model output for response. Defaults to None.
             model_format (string, optional): model format, must be "tensorflow" or "onnx" (default: "onnx" if model path ends with .onnx, "tensorflow" if model path ends with .zip or is a directory)
             tf_serving_key (string, optional): name of the signature def to use for prediction (required if your model has more than one signature def)
         

From 12645609ad956b77fe23b6e356f79fc67003ce9a Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Fri, 27 Sep 2019 17:44:33 -0400
Subject: [PATCH 19/24] Docs tweak

---
 docs/cluster/python-client.md | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/docs/cluster/python-client.md b/docs/cluster/python-client.md
index b036995e47..2aa87e645d 100644
--- a/docs/cluster/python-client.md
+++ b/docs/cluster/python-client.md
@@ -9,6 +9,8 @@ pip install git+https://github.com/cortexlabs/cortex.git@master#egg=cortex\&subd
 
 Python client needs to be initialized with AWS credentials and a url to the operator your Cortex cluster.
 
+The Operator URL can be retrieved by running `./cortex.sh endpoints`
+
 ```python
 from cortex import Client
 
@@ -29,7 +31,7 @@ api_url = cortex.deploy(
 )
 ```
 
-`api_url` is the url to the deployed API, it accepts JSON POST requests.
+`api_url` contains the url to the deployed API. It accepts JSON POST requests.
 
 ```python
 import requests

From dac09611073839bef5a83039582a2f3d5926ac40 Mon Sep 17 00:00:00 2001
From: Vishal Bollu <vishal.bollu@gmail.com>
Date: Fri, 27 Sep 2019 11:59:18 -0400
Subject: [PATCH 20/24] Prevent load balancer from timing out requests (#490)

---
 manager/manifests/istio-values.yaml | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/manager/manifests/istio-values.yaml b/manager/manifests/istio-values.yaml
index a26482e56d..6b88abb988 100644
--- a/manager/manifests/istio-values.yaml
+++ b/manager/manifests/istio-values.yaml
@@ -76,6 +76,8 @@ gateways:
       limits:
        cpu: 2000m
        memory: 1024Mi
+    serviceAnnotations:
+      service.beta.kubernetes.io/aws-load-balancer-connection-idle-timeout: 10 * 60 # 10 minutes
     type: LoadBalancer
     ports:
     - port: 80

From 0656e67302ca9eb19a07655fd70c5eb062e8db05 Mon Sep 17 00:00:00 2001
From: Omer Spillinger <42219498+ospillinger@users.noreply.github.com>
Date: Fri, 27 Sep 2019 15:20:47 -0700
Subject: [PATCH 21/24] Update python-client.md

---
 docs/cluster/python-client.md | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)

diff --git a/docs/cluster/python-client.md b/docs/cluster/python-client.md
index 2aa87e645d..d8906964d3 100644
--- a/docs/cluster/python-client.md
+++ b/docs/cluster/python-client.md
@@ -1,23 +1,21 @@
 # Python Client
 
-Python Client can be used to programmatically deploy to a Cortex Cluster.
+The Python client can be used to programmatically deploy models to a Cortex Cluster.
 
 <!-- CORTEX_VERSION_MINOR -->
 ```
 pip install git+https://github.com/cortexlabs/cortex.git@master#egg=cortex\&subdirectory=pkg/workloads/cortex/client
 ```
 
-Python client needs to be initialized with AWS credentials and a url to the operator your Cortex cluster.
-
-The Operator URL can be retrieved by running `./cortex.sh endpoints`
+The Python client needs to be initialized with AWS credentials and an operator URL for your Cortex cluster. You can find the operator URL by running `./cortex.sh endpoints`.
 
 ```python
 from cortex import Client
 
 cortex = Client(
-    aws_access_key_id="<string>",  # AWS access key associated with the account that created the Cortex cluster
-    aws_secret_access_key="<string>",  # AWS secrey key associated with the provided AWS access key  
-    operator_url="<string>" # Operator URL of your cortex cluster
+    aws_access_key_id="<string>",  # AWS access key associated with the account that the cluster is running on
+    aws_secret_access_key="<string>",  # AWS secret key associated with the AWS access key
+    operator_url="<string>" # operator URL of your cluster
 )
 
 api_url = cortex.deploy(
@@ -31,16 +29,17 @@ api_url = cortex.deploy(
 )
 ```
 
-`api_url` contains the url to the deployed API. It accepts JSON POST requests.
+`api_url` contains the URL of the deployed API. The API accepts JSON POST requests.
 
 ```python
 import requests
 
 sample = {
-    "feature_1" [0,1,2]
+  "feature_1": 'a',
+  "feature_2": 'b',
+  "feature_3": 'c'
 }
 
 resp = requests.post(api_url, json=sample)
-
-resp.json() # your model prediction
-```
\ No newline at end of file
+resp.json()
+```

From 03feb6e6a59519c08b612f909fd62a506460f3b0 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Fri, 27 Sep 2019 18:25:08 -0400
Subject: [PATCH 22/24] Fix linting

---
 pkg/workloads/cortex/client/cortex/client.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index ad5d5ca12a..2c4e99067e 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -32,10 +32,10 @@
 class Client(object):
     def __init__(self, aws_access_key_id, aws_secret_access_key, operator_url):
         """Initialize a Client to a Cortex Operator
-        
+
         Args:
             aws_access_key_id (string): AWS access key associated with the account that created the Cortex Cluster
-            aws_secret_access_key (string): AWS secrey key associated with the provided AWS access key  
+            aws_secret_access_key (string): AWS secrey key associated with the provided AWS access key
             operator_url (string): Operator URL of your cortex cluster
         """
 
@@ -63,16 +63,16 @@ def deploy(
         tf_serving_key=None,
     ):
         """Deploy an API
-        
+
         Args:
             deployment_name (string): deployment name
             api_name (string): API name
             model_path (string): S3 path to an exported model
-            pre_inference (function, optional): function used to prepare requests for model input. Defaults to None.
-            post_inference (function, optional): function used to prepare model output for response. Defaults to None.
+            pre_inference (function, optional): function used to prepare requests for model input
+            post_inference (function, optional): function used to prepare model output for response
             model_format (string, optional): model format, must be "tensorflow" or "onnx" (default: "onnx" if model path ends with .onnx, "tensorflow" if model path ends with .zip or is a directory)
             tf_serving_key (string, optional): name of the signature def to use for prediction (required if your model has more than one signature def)
-        
+
         Returns:
             string: url to the deployed API
         """

From a91a662e6a32989b52f3335e76ce68e4f827c388 Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Mon, 30 Sep 2019 19:18:57 +0000
Subject: [PATCH 23/24] Line up comments

---
 pkg/workloads/cortex/client/cortex/client.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 2c4e99067e..8da70c0378 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -34,9 +34,9 @@ def __init__(self, aws_access_key_id, aws_secret_access_key, operator_url):
         """Initialize a Client to a Cortex Operator
 
         Args:
-            aws_access_key_id (string): AWS access key associated with the account that created the Cortex Cluster
-            aws_secret_access_key (string): AWS secrey key associated with the provided AWS access key
-            operator_url (string): Operator URL of your cortex cluster
+            aws_access_key_id (string): AWS access key associated with the account that the cluster is running on
+            aws_secret_access_key (string): AWS secret key associated with the AWS access key
+            operator_url (string): operator URL of your cluster
         """
 
         self.operator_url = operator_url

From b74e425aee117fdb12157b162f6ba1cf9f5207db Mon Sep 17 00:00:00 2001
From: vishal <vishalbollu@users.noreply.github.com>
Date: Mon, 30 Sep 2019 15:31:47 -0400
Subject: [PATCH 24/24] Do not add none params

---
 pkg/workloads/cortex/client/cortex/client.py | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/pkg/workloads/cortex/client/cortex/client.py b/pkg/workloads/cortex/client/cortex/client.py
index 8da70c0378..3b5058cb5a 100644
--- a/pkg/workloads/cortex/client/cortex/client.py
+++ b/pkg/workloads/cortex/client/cortex/client.py
@@ -81,13 +81,13 @@ def deploy(
         api_working_dir = os.path.join(working_dir, api_name)
         pathlib.Path(api_working_dir).mkdir(parents=True, exist_ok=True)
 
-        api_config = {
-            "kind": "api",
-            "model": model_path,
-            "name": api_name,
-            "model_format": model_format,
-            "tf_serving_key": tf_serving_key,
-        }
+        api_config = {"kind": "api", "model": model_path, "name": api_name}
+
+        if tf_serving_key is not None:
+            api_config["model_format"] = tf_serving_key
+
+        if model_format is not None:
+            api_config["model_format"] = model_format
 
         if pre_inference is not None or post_inference is not None:
             reqs = subprocess.check_output([sys.executable, "-m", "pip", "freeze"])