What Is ModelOps and How Can It Help Optimize Your Enterprise Data?

What Is ModelOps and How Does It Work?

ModelOps is an umbrella term that includes tools that allow organizations to derive greater value from their AI models, explains Terry Halvorsen, vice president of federal client development at IBM. This can include DevOps, along with DataOps, ITOps and MLOps, which deals with machine learning (ML).

Notably, ModelOps also involves tools related to data management and data cleaning. And ideally, these tools will leverage automation. “One of the big problems with all of this — and implementing enterprise AI and cleaning up your data — is that there aren’t enough skilled people to do the work,” Halvorsen says.

“So, how do I use these techniques to automate things and reduce my requirement for data scientists?” he asks. “I could do that with this set of tools and processes.”

ModelOps focuses on the governance of the full lifecycle of ML models and ensures that these models are updated when they start to become less stable and lose their predictive value, says Jennifer Atlas, director of global presales at Minitab, a data analytics firm that offers ModelOps services.

READ MORE: Why midsize IT leaders are turning cloud optimization into a competitive edge.

“From a big-picture standpoint, its job is to make sure that the model is good, holding its own, and alerting the data scientists and other people who are using that model,” Atlas says.

Enterprises can also use ModelOps to swap in new models when their main model requires fine-tuning or replacement. Atlas says that this capability ensures that models are not using biased data that will lead to biased outcomes.

What Is the Difference Between ModelOps and MLOps?

A subset of ModelOps, MLOps is a set of tools focused more on enabling data scientists and others they work with to collaborate and communicate when automating or adjusting ML models, Atlas explains. It is concerned with testing ML models and ensuring that the algorithms produce accurate results.

MLOps is also more narrowly focused on factors such as the costs of data engineering and model training. Additionally, it focuses on checking the data feeding into the ML model.

“It’s the part that makes it worthwhile, if ModelOps can be successful,” says Atlas.

Enterprise ModelOps Use Cases

There are many enterprise use cases for ModelOps across industries.

For starters, ModelOps is helpful in the context of enterprise data management and improving data quality, as organizations are inundated with wide-ranging data that they must clean, organize and derive actionable business insights from. This is particularly instrumental as 41% of business leaders lack an understanding of data because it’s either too complex or inaccessible, and 33% of business leaders cite an inability to generate insights from data, according to Salesforce.

As IBM details, ModelOps can also be useful within AI governance processes, helping provide a comprehensive plan for tracking ML assets in an enterprise. This is often accomplished through the creation of a centralized repository of fact sheets that track the lifecycle of a model — one way organizations such as Golden Bank have leveraged ModelOps to bolster their AI governance.

ModelOps can also be useful for enterprises researching where and how social initiatives can be most effective, and for those performing operational tests and other experiments for products in development — even when this research and testing is conducted outside of a traditional office building.

Atlas says that some organizations use ML models to analyze drone footage and other surveillance imagery to detect changes from previous observations. Automating that through ModelOps can be helpful for organizations performing observations in the field and analyzing the data.

ModelOps also helps organizations assess whether the data they collect and use for models is current enough for the desired application. “If I’m targeting, it better be current data and not something based on a geographic survey from three years ago,” says Halvorsen.

Halvorsen says that ModelOps can determine data viability — which refers to the shelf life of data, or how long it can be stored and still useful — as well. Plus, ModelOps can account for whether the volume of data being used for a model will limit the impact of any errors within it, while the opposite may be true if an organization trains a model on a smaller but more accurate data set.

DIVE DEEPER: How enterprises can build a GenAI strategy for long- and short-term business goals.

How Can Enterprises Realize the Benefits of ModelOps?

ModelOps is not particularly difficult to implement, but it often fails when IT leaders don’t invest enough in cleaning their data.

“People don’t have a good understanding of their data, and they frankly don’t want to pay to restructure and, in some cases, rearchitect the data to make it more valuable for use in an AI development,” Halvorsen says.

Atlas explains that for organizations to reap the benefits of ModelOps, there must be strong partnerships and communication among data scientists, engineers, IT security teams and other technologists. “The tricky part of ModelOps is that you are usually crossing a couple of different departments that have competing priorities, but it behooves them to work together to get that benefit.”

She goes on to say that it helps to have a set leader — such as a chief data officer — to bridge those gaps and ensure teams are all working together.

Toward this end, Halvorsen notes, the title of the convening figure — be it a CIO, chief data officer or chief AI officer — shouldn’t be the determining factor in deciding who leads. What’s important is that the leader is someone who “has the money and the authority to follow the fixes” and implement needed changes in the data or models. Considering the current state of budgeting, that will likely continue to be CIOs

With the needed money and authority in place, the leader should focus on how much data their enterprise is using. They must also ensure that the organization gets the most out of its data if it’s determined that large amounts aren’t fully leveraged when it comes to training models.

“You have to increase the value you’re getting from your data and how much data you’re using, but you also have to make sure that data is high quality,” Halvorsen says. “Sometimes this is hard because people want results faster.”

“You have to resist the urge to say, ‘OK, well, I’ll just throw the AI at my current data,’” he says. “That will give you crappy results.”