Engineering careers: Making AI accessible

More and more, working smarter and faster means building more intelligent software with a focus on AI capabilities to enhance machine learning, increase automation, and produce faster, more accurate predictions.

Unlocking the benefits of artificial intelligence is a central focus for the ServiceNow Advanced Technology Group (ATG)—a customer-centered innovation group founded in August 2020. It has nearly 250 employees located in Montreal, Quebec; Toronto, Ontario; Santa Clara, California; and Hyderabad, India.

[We’re hiring. Explore ServiceNow engineering careers.]

“What attracted me is the idea of developing AI and making it accessible to companies and individuals who otherwise don’t have the setup, resources, or expertise to use it,” says Parmida A., an applied research scientist in ATG.

Valérie B., ATG director of research and scientific programs, says there’s currently “a lot of thirst” for what the group can bring to the table. “It’s a matter of figuring out how to make complex things simple for people who are not from the field,” she says.

ATG’s biggest attribute, Valérie says, is knowing what’s coming five years from now—and figuring out how to integrate it into products.

For 2021, ATG’s objectives include:

• Achieving an impactful market presence for ServiceNow as a credible AI player
  
• Achieving customer success for AI with the workflows
  
• Making the Now Platform viable for market-leading AI
  
• Achieving an impactful engagement model within ServiceNow and delivering innovation to our customers
  
  

AI technologies tell the story

One way ATG is making AI accessible is through BAyesian Active Learning (BaaL). It’s currently open source and available on GitHub.

In AI, tons of labeled data are needed in order to train a machine learning model and enable it to make predictions. This labeling process might include hundreds of thousands of samples and is very expensive. This prevents many organizations from dipping their toes into the AI waters.

BaaL queries only what it deems the most effective samples for training the model rather than labeling random selections. This makes the whole procedure more efficient, smarter and, ultimately, less expensive.

Parmida says it’s a win-win for both end users and developers. “BaaL picks up patterns in the data ahead of time—flagging the end user as to which classes are in need of more data or which have more noise,” she says. “For developers, instead of digging into parameter tuning and better training, they can use BaaL to train more efficiently.”

Another innovative technology from ATG is called document intelligence, or “doc intel.” Designed for markets that have traditionally required a slew of paper documents, such as insurance companies, Doc Intel uses AI optimal character recognition (OCR) to scan large volumes of documents electronically and detect the text in question, eliminating the need for manual data entry and location. This frees resources for high-value work.

“Doc Intel allows us to use emerging, AI technologies to deliver the paperless office to industries that was promised 20 years ago,” says Daniel T., an AI developer in ATG engineering. “It’s very exciting to see the latest and greatest stuff in OCR and how it’s solving problems and making us more efficient.”

ATG is focused on creating compelling AI experiences that power intelligent workflow solutions. These are based on six areas:

• Language interaction
  
• Case intelligence
  
• Planning
  
• AI workbench
  
• Automation and discovery
  
• Data as a service (DaaS)
  
  

The importance of collaboration

The Doc Intel team is comprised of folks from the vision team, the product management team, and the team responsible for Nagini—the initiative that allows the use of Python language-based production (ServiceNow uses Java).

“In the end, every team has to make sure they understand what every other team is doing to ensure we succeed in delivering the best AI capabilities,” Daniel says.

Valérie, who leads a group of around 30 employees, says collaboration is vital so that researchers in different areas are aware of what’s going on in all related fields. “They must know what’s feasible, what direction things are going, and where the competition is,” she says.

ATG is well-equipped for that. “Our research group is very well known in the field—each working on different fields within AI,” Parmida says. “We have experts in time series, computer vision, active learning, natural language processing, to name a few.”

The road ahead

How does ATG stay ahead in the AI game? Keys include continuing to identify future advances and trends, retaining close contact with the academic communities, pushing boundaries, and attracting and retaining the best talent.

“It’s important to note that AI is not like developing traditional software,” Valérie adds. “It comes as the result of experimentation—and we don’t necessarily know what’s going to happen when we experiment. We don’t have all the answers but have the people and resources to figure it out. That’s what makes it exciting.”

Explore engineering career opportunities at ServiceNow.

ServiceNow, the ServiceNow logo, Now, and other ServiceNow marks are trademarks and/or registered trademarks of ServiceNow, Inc. in the United States and/or other countries. Other company names, product names, and logos may be trademarks of the respective companies with which they are associated.