Open Data Group Rebrands as ModelOp, Reflecting Focus on AI Model Operations

The Open Data Group (ODG) this week announced it's re-launching as ModelOp, reflecting the importance of modeling to artificial intelligence (AI) and machine learning (ML) implmentations.

Company executives view the goal of making data science pay off in real-world applications as being crucial to organizations, hence the focus on Model Operations and the name change. Failure to implement models created by data scientists is seen as a roadblock to advancing AI and ML as money-making tools.

The ModelOp press release quotes research firm Gartner Inc.'s take on the current state of modeling: "The democratization of ML techniques in the last few years has seen the proliferation of model development practices but unfortunately a majority of these models are neither operationalized nor deployed at scale. This capability is becoming critical for the survival of data science teams and this urgency will push MLOps toward the Plateau of Productivity in two to five years."

In an interview with ADTmag, ModelOp CEO Pete Foley said this is a problem he and his colleagues at the company have seen firsthand when they visit new clients. "Our first question is, do you have a challenge getting your models deployed? Do you face a challenge of getting your models into your business? And every single data scientist says, absolutely. And we come in with a CIO or CTO and say, are you challenged supporting your data scientists or your model developers? Again, absolutely."

Too often the models data scientists have created are "thrown over the wall" and beyond some scripting tweaks, the model lies dormant, Foley explained, resulting in lost productivity and profitability.

ModelOp addresses this problem with a combination of proprietary technology and consulting services, he said.

"We've had some extremely good early success with some big financial services, the big banks," Foley said. "And I think one of the reasons is that they run their businesses on models. They get it. They want to go right to Tell me how you solve this."

The heart of what Foley's company gives clients to solve the problem is proprietary technology called ModelOp Center. "It gives our customers a single pane of glass using our services and our proprietary technology in addition to all the other technical stacks that our customer will have in place already today."

The models and other work that data scientists have done previously at the company isn't lost or replaced in the ModelOp solution, he explained. The company takes an open holistic approach to helping the customer realize the value in what they have and what can be added to get the models to the point that they are revenue producers.

When calling on a new customer, Foley said: "We walk them through how we helped a large bank get a trading model for one of their desks. We created the model, all the way down, this transformational process of working with their data scientists and utilizing some modern machine learning techniques, through understanding the correct and most strategic data feeds."

This resulted in the creation of a dashboard for business users, in this case traders, so they can see the value of the model. "They don't understand how it got there but they're making trading decisions on that model," Foley said.

ModelOp called models for Data Science and Machine Learning a new kind of software in its announcement. "DS/ML models are the engines that organizations use to turn their data into value. In order to capture that value, DS/ML models must be integrated with enterprise applications that use the models' predictions to automate and improve decisions like credit scoring, fraud detection, bond trading, customer retention, manufacturing operations, supply chain optimization, ecommerce sales, ad conversions and virtually anything that can be driven by data."

Success is possible if a company is willing to invest in modeling to get to the point where ML are fully operational across the organization, Foley said.

More information can be found in an Oct. 16 blog post.