Automatic Win: Using AI to Process Data Faster
Northrop Grumman applies artificial intelligence to telemetry data on spacecraft health

By Michelle A. Monroe
Data analysis that would take a person 40 hours has been reduced to five seconds flat, thanks to a Northrop Grumman team using artificial intelligence (AI) to process telemetry data from satellites faster than humanly possible. This is a breakthrough poised to transform satellite health monitoring.
Telemetry — the process of collecting and transmitting data from remote sources for monitoring and analysis — produces massive amounts of data which systems engineers at ground stations manually monitor to determine the health and safety of a spacecraft. Northrop Grumman is completing this work faster and more efficiently with a new framework that aims to reduce processing time by 80%, resulting in millions of dollars in savings.
A New Support Tool
“With spacecraft telemetry, there’s so much data that it’s not humanly possible to intricately review everything coming in,” said James Skrinska, a Northrop Grumman engineer who leads the INFRARED/JANUS program. “We’re developing an AI solution that helps systems engineers minimize time spent on routine tasks so they can focus their talents on challenging data analysis that moves the needle for our customers.”
This first-of-its kind application of AI to systems engineering is already decreasing cost, gaining efficiency and delivering never-before-seen processing rates. The system, called INFRARED/JANUS, has been connected to several satellites since June 2024, providing access to more data in less time than ever before. Currently, Northrop Grumman has two models up and running, and the team is continuously adapting AI models onto customer products to deliver results that support end users at ground stations — all in alignment with the company’s responsible AI principles.
“We’re answering the industry’s call and the customer’s desire to move toward autonomy,” James said. “We’ve built a system that allows end users to train AI themselves. The most exciting part of this project is getting AI models in the hands of regular engineers to process the data they’re most familiar with.”
Building Trust
Imagine being a security guard watching camera surveillance of a building. A human can look at a single camera with close to 100% accuracy. But an AI program can process 100 feeds with similar accuracy — and only flag what matters.
In telemetry terms, this shifts this engineer’s role from watching raw data streams to reviewing only what AI flags.
“This is like when Excel showed up in the 1980s,” James said. “There were people who preferred graph paper and pencil — which, yes, still work, but Excel is better, faster and more efficient. It’s definitely a change in mindset when incorporating new technology into daily tasks.”
Digital transformation is both a technological and cultural shift that requires changing hearts and minds. Northrop Grumman is helping engineers embrace that change by building confidence in AI’s reliability.
“We’ve been working on an evolving AI solution connected to telemetry processing,” James said. “We’ve achieved our goal of stamping out a process and concept of operations about telemetry processing with the people on the ground. Now we’re measuring the accuracy and building trust with the people managing it.”