When you build a Machine Learning (ML) product, consider at least two MLOps scenarios. First, the model is replaceable, as breakthrough algorithms are introduced in academia or industry. Second, the model itself has to evolve with the data in the changing world.

We can handle both scenarios with the services provided by Vertex AI. For example:

• AutoML capability automatically identifies the best model based on your budget, data, and settings.
• You can easily manage the dataset with Vertex Managed Datasets by creating a new dataset or adding data to an existing dataset.
• You can build an ML pipeline to automate a series of steps that start with importing a dataset and end with deploying a model using Vertex Pipelines.

This blog post shows you how to build this system. You can find the full notebook for reproduction here. Many folks focus on the ML pipeline when it comes to MLOps, but there are more parts to building MLOps as a “system”. In this post, you will see how Google Cloud Storage (GCS) and Google Cloud Functions manage data and handle events in the MLOps system.

Architecture

Figure 1 shows the overall architecture presented in this blog. We cover the components and their connection in the context of two common workflows of the MLOps system.

Components

Vertex AI is at the heart of this system, and it leverages Vertex Managed Datasets, AutoML, Predictions, and Pipelines. We can create and manage a dataset as it grows using Vertex Managed Datasets. Vertex AutoML selects the best model without your knowing much about modeling. Vertex Predictions creates an endpoint (RestAPI) to which the client communicates.

It is a simple, fully managed yet somewhat complete end-to-end MLOps workflow moves from a dataset to training a model that gets deployed. This workflow can be programmatically written in Vertex Pipelines. Vertex Pipelines outputs the specification for an ML pipeline allowing you to re-run the pipeline whenever or wherever you want. Specify when and how to trigger the pipeline using Cloud Functions and Cloud Storage.

Cloud Functions is a serverless way to deploy your code in Google Cloud. In this particular project, it triggers the pipeline by listening to changes on the specified Cloud Storage location. Specifically, if a new dataset is added, for example, a new span number is created; the pipeline is triggered to train the dataset, and a new model is deployed.

Workflow

This MLOps system prepares the dataset with either Vertex Dataset’s built-in user interface (UI) or any external tools based on your preference. You can upload the prepared dataset into the designated GCS bucket with a new folder named SPAN-NUMBER. Cloud Functions then detects the changes in the GCS bucket and triggers the Vertex Pipeline to run the jobs from AutoML training to endpoint deployment.

Inside the Vertex Pipeline, it checks if there is an existing dataset created previously. If the dataset is new, Vertex Pipeline creates a new Vertex Dataset by importing the dataset from the GCS location and emits the corresponding Artifact. Otherwise, it adds the additional dataset to the existing Vertex Dataset and emits an artifact.

When the Vertex Pipeline recognizes the dataset as a new one, it trains a new AutoML model and deploys it by creating a new endpoint. If the dataset isn’t new, it tries to retrieve the model ID from Vertex Model and determines whether a new AutoML model or an updated AutoML model is needed. The second branch determines whether the AutoML model has been created. If it hasn’t been created, the second branch creates a new model. Also, when the model is trained, the corresponding component emits the artifact as well.

Directory structure that reflects different distributions

In this project, I have created two subsets of the CIFAR-10 dataset, SPAN-1 and SPAN-2. A more general version of this project can be found here, which shows how to build training and batch evaluation pipelines pipelines. The pipelines can be set up to cooperate so they can evaluate the currently deployed model and trigger the retraining process.

ML Pipeline with Kubeflow Pipelines (KFP)

We chose to use Kubeflow Pipelines to orchestrate the pipeline. There are a few things that I would like to highlight. First, it’s good to know how to make branches with conditional statements in KFP. Second, you need to explore AutoML API specifications to fully leverage AutoML capabilities, such as training a model based on the previously trained one. Last, you also need to find a way to emit artifacts for Vertex Dataset and Vertex Model to consume that Vertex AI can recognize them. Let’s go through these one by one.

Branching strategy

In this project, there are two main conditions and two sub-branches inside the second main branch. The main branches split the pipeline based on a condition if there is an existing Vertex Dataset. The sub-branches are applied in the second main branch, which is selected when there is an exciting Vertex Dataset. It looks up the list of models and decides to train an AutoML model from scratch or a previously trained one.

ML pipelines written in KFP can have conditions with a special syntax of kfp.dsl.Condition. For instance, we can define the branches as follows:

from google_cloud_pipeline_components import aiplatform as gcc_aip


# try to get Vertex Dataset ID
dataset_op = get_dataset_id(...) 

with kfp.dsl.Condition(name="create dataset", 
                       dataset_op.outputs['Output'] == 'None'):
    # Create Vertex Dataset, train AutoML from scratch, deploy model

with kfp.dsl.Condition(name="update dataset", 
                       dataset_op.outputs['Output'] != 'None'):
    # Update existing Vertex Dataset
    ...

    # try to get Vertex Model ID
    model_op = get_model_id(...)

    with kfp.dsl.Condition(name='model not exist',
                           model_op.outputs['Output'] == 'None'):
    # Create Vertex Dataset, train AutoML from scratch, deploy model

    with kfp.dsl.Condition(name='model exist',
                           model_op.outputs['Output'] != 'None'):
        # Create Vertex Dataset, train AutoML based on trained one, deploy model

get_dataset_id and get_model_id are custom KFP components used to determine if there is an existing Vertex Dataset and Vertex Model respectively. Both return “None” if a model is found and some other value if a model isn’t found. They also emit Vertex AI-aware artifacts. You will see what this means in the next section.

Emit Vertex AI-aware artifacts

Artifacts track the path of each experiment in the ML pipeline and display metadata in the Vertex Pipeline UI. When Vertex AI aware artifacts are released into in the pipeline, Vertex Pipeline UI displays links for its internal services such as Vertex Dataset, so that users can visit a web page for more information.

So how could you write a custom component to generate Vertex AI-aware artifacts? To do this, custom components should have Output[Artifact] in their parameters. Then you need to replace the resourceName of the metadata attribute with a special string format.

The following code example is the actual definition of get_dataset_id used in the previous code snippet:

@component(
    packages_to_install=["google-cloud-aiplatform", 
                         "google-cloud-pipeline-components"]
)
def get_dataset_id(project_id: str, 
                     location: str,
                 dataset_name: str,
                 dataset_path: str,
                      dataset: Output[Artifact]) -> str:
    from google.cloud import aiplatform
    from google.cloud.aiplatform.datasets.image_dataset import ImageDataset
    from google_cloud_pipeline_components.types.artifact_types import VertexDataset

    
    aiplatform.init(project=project_id, location=location)
    
    datasets = aiplatform.ImageDataset.list(project=project_id,
                                            location=location,
                                            filter=f'display_name={dataset_name}')
    
    if len(datasets) > 0:
        dataset.metadata['resourceName'] = 
               f'projects/{project_id}/locations/{location}/datasets/{datasets[0].name}'
        return f'projects/{project_id}/locations/{location}/datasets/{datasets[0].name}'
    else:
        return 'None'

As you see, the dataset is defined in the parameters as Output[Artifact]. Even though it appears in the parameter, it is actually emitted automatically. You just need to provide the necessary data as if it is a function variable.

The dataset component retrieves the list of Vertex Dataset by calling the aiplotform.ImageDataset.list API. If the length of it is zero, it simply returns ‘None’. Otherwise, it returns the found resource name of the Vertex Dataset and provides the dataset.metadata[‘resourceName’] with the resource name at the same time. The Vertex AI-aware resource name follows a special string format, which is ‘projects/<project-id>/locations/<location>/<vertex-resource-type>/<resource-name>’.

The <vertex-resource-type>can be anything that points to an internal Vertex AI service. For instance, if you want to specify that the artifact is the Vertex Model, then you should replace <vertex-resource-type> with models. The <resource-name> is the unique ID of the resource, and it can be accessed in the name attribute of the resource found by the aiplatform API. The other custom component, get_model_id, is written in a very similar way as well.

AutoML based on the previous model

You sometimes want to train a new model on top of the previously best model. If that is possible, the new model will probably be much better than the one trained from scratch, because it leverages previously learned knowledge.

Luckily, Vertex AutoML comes with the ability to train a model using a previous model. AutoMLImageTrainingJobRunOp component lets you train a model by simply providing the base_model argument as follows:

training_job_run_op =
gcc_aip.AutoMLImageTrainingJobRunOp(
…,
base_model=model_op.outputs['model'],
…
)

When training a new AutoML model from scratch, you pass ‘None‘ in the base_model argument, and it is the default value. However, you can set it with a VertexModel artifact, and the component will trigger an AutoML training job based on the other model.

One thing to be careful of is that VertexModel artifacts can’t be constructed in a typical way of Python programming That means you can’t create an instance of VertexModel artifact by setting the id found in the Vertex Model dashboard. The only way you can create one is to set the metadata[‘resourceName’] parameters properly. The same rule applies to other Vertex AI-related artifacts such as VertexDataset. You can see how the VertexDataset artifact is constructed properly to get an existing Vertex Dataset to import additional data into it. See the full notebook of this project here.

Cost

You can reproduce the same result from this project with the free $300 credit when you create a new GCP account.

At the time of this blog post, Vertex Pipelines costs about $0.03/run, and the type of underlying VM for each pipeline component is e2-standard-4, which costs about $0.134/hour. Vertex AutoML training costs about $3.465/hour for image classification. GCS holds the actual data, which costs about $2.40/month for 100GiB capacity, and Vertex Dataset is free.

To simulate two different branches, the entire experiment took about one to two hours, and the total cost for this project is approximately $16.59. Please find more detailed pricing information about Vertex AI here.

Conclusion

Many people underestimate the capability of AutoML, but it is a great alternative for app and service developers who have little ML background. Vertex AI is a great platform that provides AutoML as well as Pipeline features to automate the ML workflow. In this article, I have demonstrated how to set up and run a basic MLOps workflow, from data injection to training a model based on the previously-achieved best one, to deploying the model to a Vertex AI platform. With this, we can let our ML model automatically adapt to the changes in a new dataset. What’s left for you to implement is to integrate a model monitoring system to detect data/model drift. One example is found here.

With a growing fleet of 325 aircraft that cover more than 1,000 routes across 158 airports, easyJet is one of Europe’s most popular airlines. And easyJet serves an average of 90 million passengers each year, so a helpful mobile experience for its customers is a top priority.

Travellers today are inherently mobile-first, so finding new ways to make it easier for them to search and book flights is key. To do exactly that, easyJet partnered with technology company Travelport to develop Speak Now, a new feature on easyJet’s mobile app that interprets voice searches to deliver accurate and relevant flight information to travelers.

See how it works:

Powered by Dialogflow, Google Cloud’s natural language understanding tool for building conversational experiences, Speak Now lets customers ask questions to determine exactly what they’re looking for—from destinations, to dates and times, to airports they want to fly from. 

How do we create conversational experiences across devices and platforms for enterprises?
It’s clear that the rise in voice search is changing the way we go about our daily lives. Twenty-seven percent of the global online population already uses voice search on mobile, and the rapid adoption of this technology is reshaping entire industries. It’s no surprise that easyJet looked to adopt this technology to positively transform experiences for their customers.   

Dialogflow, a core component of Google Cloud Contact Center AI, makes it easy to build accurate, flexible conversation interfaces that allow users to ask questions and accomplish tasks in everyday language. It understands the nuances of human language and translates end-user text or audio during a conversation to structured data that apps and services can understand. 

Speak Now is a great example of how we’re using cloud technologies and AI to make the experience of buying and managing travel continually better for everyone.

Daniel Young, Head of Digital Experience, easyJet

Daniel Young, Head of Digital Experience at easyJet commented: “We picked Dialogflow due to its strengths and ease with which a powerful conversational agent can be built. Speak Now is a great example of how we’re using cloud technologies and AI to make the experience of buying and managing travel continually better for everyone. This is the latest in a series of innovative features that will make booking travel as easy as it can possibly be, giving easyJet customers a helpful digital experience.”

Consumers already rely on voice assistants to play their favorite music, add items to a shopping list, and order taxis, Speak Now is a great example of how voice assistants can now make the customer experience better and more intuitive for travel.

For a business to succeed in the long term, it needs to learn not just to adapt to inevitable change, but to harness it. South Africa-based PPS has been an insurance company since 1941 and today is the biggest mutual insurance provider in the country.

As a mutual company, PPS is owned by more than 200,000 members, making them shareholders. In recent years, PPS and other companies like it have been affected by a number of external factors.

“For one thing, technology platforms have brought in a new gig economy that has all kinds of implications for insurance,” says Avsharn Bachoo, CTO at PPS. “What we’ve been seeing is basically a disruption of the South African insurance industry. We chose to see that as an opportunity.”

“Our servers were at the end of their life cycle and we had to decide whether to refresh them or switch completely. To embrace the world of AI and machine learning effectively, we knew we needed a cloud-based infrastructure. We’ve found the answer in Google Cloud Platform.”

—Avsharn Bachoo, CTO, PPS

In early 2018, faced with an uncertain economic environment that was squeezing growth and profitability, PPS decided to transform itself from a traditional broker-based business into a digital insurance provider. A key pillar of this new strategy was to overhaul the company’s technology infrastructure. To turn the strategy into reality, Avsharn and his team chose Google Cloud Platform (GCP).

“Our servers were at the end of their life cycle and we had to decide whether to refresh them or switch completely,” says Avsharn. “To embrace the world of AI and machine learning (ML) effectively, we knew we needed a cloud-based infrastructure. We’ve found the answer in Google Cloud Platform.”

Power, speed, flexibility with Google Cloud Platform

Previously, PPS maintained an on-premises IT infrastructure, which worked for its traditional business but was unsuited for its new way of working. In early 2018, the company started working on new products for its members but this required large amounts of compute power that proved prohibitively expensive with on-premises servers. Even existing products were starting to require more than the infrastructure could deliver. Aging equipment meant that it’s testing and quality assurance environments bore little resemblance to the actual production environment.

“We had no pre-production environments at all,” says Avsharn, resulting in more work for developers after products had been released. Meanwhile, the capital required to buy and configure more servers for new projects meant fewer resources available for innovation, and left the company less able to react to changes in the market. PPS knew it had to find a cloud-based alternative.

Shortly after devising a new digital strategy, PPS engineers attended a training session on cloud infrastructure given by leading South African Google Cloud Partner Siatik. Impressed with the presentation, PPS engaged Siatik to help run a proof of concept for a cloud-based infrastructure, running on GCP. With on-site engineers and constant communication, Siatik formed a very close working relationship with PPS. “The team at Siatik was exemplary,” recalls Avsharn. “They were well-organized, with cutting-edge technical acumen and very creative solutions to our problems. They were real game-changers.”

“We wanted the platform to retrain its models in response to new data and improve its recommendations with more information. Normally this would be a manual process but Google Cloud ML Engine lets the models do this automatically.”

—Kimoon Kim, Lead Solution Architect and Data Engineer, Siatik

The proof of concept was successful, with GCP outperforming the existing infrastructure in terms of how it handled compute demands, databases, and storage.

“It’s the speed of GCP that really impresses us,” says Avsharn. PPS saw that GCP wasn’t just an opportunity to migrate its existing infrastructure to the cloud. With Siatik’s help, it redesigned its monolithic core architecture to one based around microservices using Google Kubernetes Engine (GKE). For data processing and storage, Cloud Dataflow and Cloud Datastore proved invaluable, while Stackdriver helped the IT team stay on top of logging and monitoring the system.

“Google Cloud makes migrations very easy,” says Brett St. Clair, CEO at Siatik. “It takes care of all the hard work with configurations and replications, so when we switch the machines on, everything is ready and working.”

The ease with which PPS migrated to GCP means that it can now tackle strategic goals much more quickly than before. The most ambitious of these is an AI-powered product recommendation platform. Information is collected from customers who opt in at a defined point in their journey, this database is queried using BigQuery, and the information is fed into the platform. The AI model then calculates the most appropriate products for each member, according to their personal history.

“Most of the product recommendation engines out there are based on clustering, where you’re offered products based on your peer groups,” explains Avsharn. “For the first time, we can make recommendations to members based on their individual preferences and historical behavior. That’s really powerful for us.”

Siatik helped PPS use TensorFlow and Cloud Machine Learning Engine to build the AI platform. For the engineers, these easy-to-use tools helped speed up the process considerably, allowing them to host the models locally without any fuss. Previously, it took one to three months to manually build the model and match an offer to a customer. With the AI platform, a match takes just a few minutes. Cloud ML Engine, in particular, helped the platform adapt to new information on the fly and easily make adjustments to its hyperparameters, that is, preset variables which define the model-training process.

“We wanted the platform to retrain its models in response to new data and improve its recommendations with more information,” says Kimoon Kim, Lead Solution Architect and Data Engineer at Siatik. “Normally this would be a manual process but Google Cloud ML Engine lets the models do this automatically.”

“Google Cloud helped us cancel out a lot of the noise around machine learning and AI. We don’t have to build new complicated algorithms or hire huge teams of data scientists to benefit. We just bring our data and use the right tools to focus on what’s really important.”

—Avsharn Bachoo, CTO, PPS

Harnessing artificial intelligence for real-world results

PPS deployed its new AI recommendation platform in December, 2018. Just a couple of months later, its impact was clear. “In around eight weeks, we saw a 5 percent growth in sales,” says Avsharn. “It’s been a direct result of building our recommendation platform with Google Cloud. We can offer the right products to the right members.”

For developers and engineers at PPS, working with Google Cloud gives them access to high performance technology and automation options with GKE. As a result, the infrastructure runs 70 percent faster than before with fewer cores and less memory. Developers can also work in mature testing environments, and for the first time, are able to build pre-production environments, leading to better quality products. More strategically, moving to a serverless, cloud-based infrastructure has helped PPS take control of its budget, moving away from intermittent, large capital spends to more manageable, project-to-project flows of operational expenditure. The company expects to see savings of around 50 percent, or $695,000.

“We have a lot more flexibility with our resources thanks to Google Cloud,” says Avsharn. “When we have a new idea, we don’t have to outlay new capital such as servers before we can even start working on it. We just spin up instances when we want and spin them back down when we’re done.”

With the AI platform deployed and working well, PPS is already looking at ways to improve it, including real-time updates and further automation. Soon, the company will integrate the platform with more sales campaigns for more effective targeting to boost sales even further. Meanwhile, it’s also experimenting with machine learning to spot patterns in data at scale for fraud analytics and risk assessment.

For PPS, working with Google Cloud has helped it transform quickly and effectively from disrupted to disruptor. The company is now looking to gain the same transformative effects by implementing G Suite for increased productivity and collaboration.

“Google Cloud helped us cancel out a lot of the noise around machine learning and AI,” says Avsharn. “We don’t have to build new complicated algorithms or hire huge teams of data scientists to benefit. We just bring our data and use the right tools to focus on what’s really important.”