Merge branch 'feature/docs' into develop

Author: praneet22

README.md

@@ -1,96 +1,72 @@
-### Author: Praneet Singh Solanki
+### Author: | Praneet Singh Solanki | Richin Jain |
 
-# DevOps For AI
+# DevOps for AI
 
 [![Build Status](https://dev.azure.com/customai/DevopsForAI-AML/_apis/build/status/Microsoft.DevOpsForAI?branchName=master)](https://dev.azure.com/customai/DevopsForAI-AML/_build/latest?definitionId=1&branchName=master)
 
-[DevOps for AI template](https://azuredevopsdemogenerator.azurewebsites.net/?name=azure%20machine%20learning) will help you to understand how to build the Continuous Integration and Continuous Delivery pipeline for a ML/AI project. We will be using the Azure DevOps Project for build and release pipelines along with Azure ML services for ML/AI model management and operationalization. 
 
-This template contains code and pipeline definition for a machine learning project demonstrating how to automate the end to end ML/AI project. The build pipelines include DevOps tasks for data sanity test, unit test, model training on different compute targets, model version management, model evaluation/model selection, model deployment as realtime web service, staged deployment to QA/prod, integration testing and functional testing.
+
+DevOps for AI will help you to understand how to build the Continuous Integration and Continuous Delivery pipeline for a ML/AI project. We will be using the Azure DevOps Project for build and release/deployment pipelines along with Azure ML services for model retraining pipeline, model management and operationalization. 
+
+This template contains code and pipeline definition for a machine learning project demonstrating how to automate the end to end ML/AI project. The build pipelines include DevOps tasks for data sanity test, unit test, model training on different compute targets, model version management, model evaluation/model selection, model deployment as realtime web service, staged deployment to QA/prod and integration testing.
+
 
 ## Prerequisite
 - Active Azure subscription
-- Minimum contributor access to Azure subscription
+- At least contributor access to Azure subscription
 
 ## Getting Started:
 
-### Import the DevOps for AI solution template from Azure DevOps Demo Generator: [Click here](https://azuredevopsdemogenerator.azurewebsites.net/?name=azure%20machine%20learning)
-
-Skip above step if already done.
-
-Once the template is imported for personal Azure DevOps account using DevOps demo generator, you need to follow below steps to get the pipeline running:
-
-### Update Pipeline Config:
-
-#### Build Pipeline
-1. Go to the **Pipelines -> Builds** on the newly created project and click **Edit** on top right
-![EditPipeline1](/docs/images/EditPipeline1.png)
-2. Click on **Create or Get Workspace** task, select the Azure subscription where you want to deploy and run the solution, and click **Authorize**
-![EditPipeline2](/docs/images/EditPipeline2.png)
-3. Click all other tasks below it and select the same subscription (no need to authorize again)
-4. Once the tasks are updated with subscription, click on **Save & queue** and select **Save**
-![EditPipeline3](/docs/images/EditPipeline3.png)
-
-#### Release Pipeline
-1. Go to the **Pipelines -> Releases** and click **Edit** on top  
-![EditPipeline4](/docs/images/EditPipeline4.png)
-2. Click on **1 job, 4 tasks** to open the tasks in **QA stage**
-![EditPipeline5](/docs/images/EditPipeline5.png)
-3. Update the subscription details in two tasks 
-![EditPipeline6](/docs/images/EditPipeline6.png)
-4. Click on **Tasks** on the top to switch to the Prod stage, update the subscription details for the two tasks in prod
-![EditPipeline7](/docs/images/EditPipeline7.png)
-5. Once you fix all the missing subscription, the **Save** is no longer grayed, click on save to save the changes in release pepeline
-![EditPipeline8](/docs/images/EditPipeline8.png)
-
-### Update Repo config:
-1. Go to the **Repos** on the newly created Azure DevOps project
-2. Open the config file [/aml_config/config.json](/aml_config/config.json) and edit it
-3. Put your Azure subscription ID in place of <>
-4. Change resource group and AML workspace name if you want
-5. Put the location where you want to deploy your Azure ML service workspace
-6. Save the changes and commit these changes to master branch 
-7. The commit will trigger the build pipeline to run deploying AML end to end solution
-8. Go to **Pipelines -> Builds** to see the pipeline run
-
-## Steps Performed in the Build Pipeline:
-
-1. Prepare the python environment
-2. Get or Create the workspace
-3. Submit Training job on the remote DSVM / Local Python Env
-4. Register model to workspace
-5. Create Docker Image for Scoring Webservice
-6. Copy and Publish the Artifacts to Release Pipeline
-
-## Steps Performed in the Release Pipeline
-In Release pipeline we deploy the image created from the build pipeline to Azure Container Instance and Azure Kubernetes Services
-
-### Deploy on ACI - QA Stage
-1. Prepare the python environment
-2. Create ACI and Deploy webservice image created in Build Pipeline
-3. Test the scoring image
-
-### Deploy on AKS - PreProd/Prod Stage
-1. Prepare the python environment
-2. Deploy on AKS
-    - Create AKS and create a new webservice on AKS with the scoring docker image
-
-        OR
-
-    - Get the existing AKS and update the webservice with new image created in Build Pipeline
-3. Test the scoring image
+To deploy this solution in your subscription, follow the manual instructions in the [getting started](docs/getting_started.md) doc
+
+
+## Architecture Diagram
+
+This reference architecture shows how to implement continuous integration (CI), continuous delivery (CD), and retraining pipeline for an AI application using Azure DevOps and Azure Machine Learning. The solution is built on the scikit-learn diabetes dataset but can be easily adapted for any AI scenario and other popular build systems such as Jenkins and Travis. 
+
+![Architecture](/docs/images/Architecture_DevOps_AI.png)
+
+
+## Architecture Flow
+
+1. Data Scientist writes/updates the code and push it to git repo. This triggers the Azure DevOps build pipeline (continuous integration).
+2. Once the Azure DevOps build pipeline is triggered, it runs following type of tasks:
+    - Run for new code: Every time new code is committed to the repo, build pipeline performs data sanity test and unit tests the new code.
+
+    - One-time run: These tasks runs only for the first time build pipeline run, they create [Azure ML Service Workspace](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#workspace), [Azure ML Compute](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-set-up-training-targets#amlcompute) used as model training compute and publish a [Azure ML Pipeline](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-ml-pipelines) with code. This published Azure ML pipeline is the model training/retraining pipeline.
+
+    `Note: The task Publish Azure ML pipeline currently runs for every code change`
+
+3. The Azure ML Retraining pipeline is triggered once the Azure DevOps build pipeline completes. All the tasks in this pipeline runs on Azure ML Compute created earlier. Following are the tasks in this pipeline:
+
+    - **Train Model** task executes model training script on Azure ML Compute. It outputs a [model](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#model) file which is stored in the [run history](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#run)
+
+    - **Evaluate Model** task evaluates the performance of newly trained model with the model in production. If new trained model performs better than the production model, next steps are executed. Else next steps are skipped.
+
+    - **Register Model** task takes the new trained better performing model and registers it with the [Azure ML Model registry](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#model-registry) to version control it.
+
+    - **Package Model** task packages the new trained model along with scoring file and python dependencies into a docker [image](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#image) and pushes it to [Azure Container Registry](https://docs.microsoft.com/en-us/azure/container-registry/container-registry-intro). This image is used to deploy model as [web service](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#web-service).
+    
+4. Once a new model scoring image is pushed to Azure Container Registry, Azure DevOps Release/Deployment pipeline is triggered. This pipeline deploys the model scoring image into Staging/QA and PROD environments.
+
+    - In the Staging/QA, one task creates [Azure Container Instance](https://docs.microsoft.com/en-us/azure/container-instances/container-instances-overview) and deploy scoring image as [web service](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#web-service) on it. 
+    
+    - The second task test this web service by calling its REST endpoint with dummy data.
+
+    
+5. The deployment in production is a [gated release](https://docs.microsoft.com/en-us/azure/devops/pipelines/release/approvals/gates?view=azure-devops). Which means, once the model web service deployment in Staging/QA environment is successful, a notification is sent to approvers to manually review and approve the release. Once the release is approved, the model scoring web service is deployed to [Azure Kubernetes Service(AKS)](https://docs.microsoft.com/en-us/azure/aks/intro-kubernetes) and the deployment is tested.
 
 ### Repo Details
 
-You can find the details of the code ans scripts in the repository [here](/docs/code_description.md)
+You can find the details of the code and scripts in the repository [here](/docs/code_description.md)
 
 ### References
-
 - [Azure Machine Learning(Azure ML) Service Workspace](https://docs.microsoft.com/en-us/azure/machine-learning/service/overview-what-is-azure-ml)
 
 - [Azure ML Samples](https://docs.microsoft.com/en-us/azure/machine-learning/service/samples-notebooks)
 - [Azure ML Python SDK Quickstart](https://docs.microsoft.com/en-us/azure/machine-learning/service/quickstart-create-workspace-with-python)
 - [Azure DevOps](https://docs.microsoft.com/en-us/azure/devops/?view=vsts)
+- [DevOps for AI template (Old Version)](https://azuredevopsdemogenerator.azurewebsites.net/?name=azure%20machine%20learning)
 
 # Contributing
 

azure-pipeline-yaml/simple-azure-pipelines.yml

@@ -8,6 +8,11 @@ pool:
 variables:
 - group: AzureKeyVaultSecrets
 
+trigger:
+- master
+- releases/*
+- develop
+
 steps:
 - task: UsePythonVersion@0
   inputs:

azure-pipelines.yml

@@ -23,8 +23,10 @@
 ARISING IN ANY WAY OUT OF THE USE OF THE SOFTWARE CODE, EVEN IF ADVISED OF THE
 POSSIBILITY OF SUCH DAMAGE.
 """
+import pickle
 import json
 import numpy
+from sklearn.ensemble import RandomForestClassifier
 from azureml.core.model import Model
 
 

code/scoring/score.py

@@ -23,6 +23,8 @@
 ARISING IN ANY WAY OUT OF THE USE OF THE SOFTWARE CODE, EVEN IF ADVISED OF THE
 POSSIBILITY OF SUCH DAMAGE.
 """
+import pickle
+from azureml.core import Workspace
 from azureml.core.run import Run
 import os
 import argparse
@@ -32,41 +34,69 @@
 from sklearn.model_selection import train_test_split
 from sklearn.externals import joblib
 import numpy as np
+import json
+import subprocess
+from typing import Tuple, List
+
+
+parser = argparse.ArgumentParser("train")
+parser.add_argument(
+    "--config_suffix", type=str, help="Datetime suffix for json config files"
+)
+parser.add_argument(
+    "--json_config",
+    type=str,
+    help="Directory to write all the intermediate json configs",
+)
+args = parser.parse_args()
+
+print("Argument 1: %s" % args.config_suffix)
+print("Argument 2: %s" % args.json_config)
+
+if not (args.json_config is None):
+    os.makedirs(args.json_config, exist_ok=True)
+    print("%s created" % args.json_config)
+
+run = Run.get_context()
+exp = run.experiment
+ws = run.experiment.workspace
 
-# using diabetes dataset from scikit-learn
 X, y = load_diabetes(return_X_y=True)
+columns = ["age", "gender", "bmi", "bp", "s1", "s2", "s3", "s4", "s5", "s6"]
 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)
 data = {"train": {"X": X_train, "y": y_train}, "test": {"X": X_test, "y": y_test}}
 
+print("Running train.py")
+
+# Randomly pic alpha
+alphas = np.arange(0.0, 1.0, 0.05)
+alpha = alphas[np.random.choice(alphas.shape[0], 1, replace=False)][0]
+print(alpha)
+run.log("alpha", alpha)
+reg = Ridge(alpha=alpha)
+reg.fit(data["train"]["X"], data["train"]["y"])
+preds = reg.predict(data["test"]["X"])
+run.log("mse", mean_squared_error(preds, data["test"]["y"]))
+
 
-def experiment_code(data_split):
-    run = Run.get_submitted_run()
-    # Randomly pic alpha
-    alphas = np.arange(0.0, 1.0, 0.05)
-    alpha = alphas[np.random.choice(alphas.shape[0], 1, replace=False)][0]
-    print(alpha)
-    # Log alpha metric
-    run.log("alpha", alpha)
-    # train the model with selected value of alpha and log mse
-    reg = Ridge(alpha=alpha)
-    reg.fit(data["train"]["X"], data_split["train"]["y"])
-    preds = reg.predict(data["test"]["X"])
-    run.log("mse", mean_squared_error(preds, data_split["test"]["y"]))
+# Save model as part of the run history
+model_name = "sklearn_regression_model.pkl"
+# model_name = "."
 
-    # Write model name to the config file
-    model_name = "sklearn_regression_model.pkl"
-    with open(model_name, "wb"):
-        joblib.dump(value=reg, filename=model_name)
+with open(model_name, "wb") as file:
+    joblib.dump(value=reg, filename=model_name)
 
-    # upload the model file explicitly into artifacts
-    run.upload_file(name="./outputs/" + model_name, path_or_stream=model_name)
-    print("Uploaded the model {} to experiment {}".format(model_name, run.experiment.name))
-    dirpath = os.getcwd()
-    print(dirpath)
+# upload the model file explicitly into artifacts
+run.upload_file(name="./outputs/" + model_name, path_or_stream=model_name)
+print("Uploaded the model {} to experiment {}".format(model_name, run.experiment.name))
+dirpath = os.getcwd()
+print(dirpath)
+print("Following files are uploaded ")
+print(run.get_file_names())
 
-    print("Following files are uploaded ")
-    print(run.get_file_names())
-    run.complete()
+# register the model
+# run.log_model(file_name = model_name)
+# print('Registered the model {} to run history {}'.format(model_name, run.history.name))
 
 run_id = {}
 run_id["run_id"] = run.id
@@ -76,6 +106,4 @@ def experiment_code(data_split):
 with open(output_path, "w") as outfile:
     json.dump(run_id, outfile)
 
-if __name__ == "__main__":
-    print("Running train.py")
-    experiment_code(data)
+run.complete()

code/training/train.py

@@ -0,0 +1,23 @@
+## Getting Started with this Repo
+
+### 1. Get the source code
+- Either clone the repository to your workspace and create your own repo with the code in GitHub.
+- An easier way is to just fork the project, so you have the repoitory under your username on GitHub itslef.
+
+
+### 2. Create Azure DevOps account
+We use Azure DevOps for running our build(CI), retraining trigger and release (CD) pipelines. If you don't already have Azure DevOps account, create one by following the instructions [here](https://docs.microsoft.com/en-us/azure/devops/organizations/accounts/create-organization?view=azure-devops)
+
+If you already have Azure DevOps account, create a new project.
+
+**Note:** Make sure you have the right permissions in Azure DevOps to do so.
+
+### 3. Create Service Principal to Login to Azure and create resources
+
+To create service principal, register an application entity in Azure Active Directory (Azure AD) and grant it the Contributor or Owner role of the subscription or the resource group where the web service belongs to. See [how to create service principal](https://docs.microsoft.com/en-us/azure/active-directory/develop/howto-create-service-principal-portal) and assign permissions to manage Azure resource
+
+**Note:** You must have sufficient permissions to register an application with your Azure AD tenant, and assign the application to a role in your Azure subscription. Contact your subscription adminstator if you don't have the permissions. Normally a subscription admin will create a Service principal and will provide you the details.
+
+
+
+