From Computer Science to Crop Science

By Lara Ivanitch

Steve Amerige’s path to studying crop science at NC State University began in the late 1950s at age 4 with a packet of seeds in the yard of his small Long Island home. He watched as the seeds dropped into small holes he had poked into the ground, and in time, the reward for watering and patience emerged: tiny zinnia seedlings poking their way out of the soil. Over the years, as Amerige grew, his garden grew, until it spanned one side of the 1,300-square-foot home he shared with his parents, grandmother and five siblings. 

“Ever since then, wherever I’ve gone, a garden has been part of my life,” he says.

Now, more than 60 years later, the graduate student is preparing for a second career, one that combines his 40 years in software development with his lifelong love of plants. 

Amerige began taking classes in the Department of Horticultural Science in 2024 as a part-time student while working 50-plus hours a week. Then in the spring of 2025, he left SAS Institute as a distinguished software developer to pursue a doctoral degree full-time at NC State. He also worked for nearly two decades at Adobe Systems before moving to North Carolina.

Drawn to crop science because of its use of predictive analytics, computer tools and research that lends itself to broader application, Amerige changed his focus to the Department of Crop and Soil Sciences when he went all-in, zeroing in on precision agriculture and computer vision.

Growing Innovation

Most of the promise of precision agriculture has not been realized yet, says Amerige’s advisor and professor Chris Reberg-Horton, who serves as the platform director for Resilient Agricultural Systems with the N.C. Plant Sciences Initiative.

Reberg-Horton believes computer vision (a branch of artificial intelligence) will allow precision agriculture to flourish. This type of agriculture allows farmers to act in targeted ways. Instead of treating entire fields with the same fertilizers, herbicides and insecticides, farmers can pinpoint which plants need which treatment using cutting-edge technology. With rising costs and shrinking profits, this precise approach will contribute to increasing resilience in agriculture.

“The combination of Steve’s experience in software development and passion for agriculture creates a unique student,” Reberg-Horton says. The two initially met at the N.C. Plant Sciences Initiative’s Hackathon in 2024, where Amerige’s won first place. “Right now, Steve is working with the embedded software team that is developing an open source software stack for building AI camera systems.”

Why the focus on cameras, specifically? Amerige explains that although drones and cameras that mount to farm equipment have grown affordable for many farmers, the data they capture often requires substantial processing and model inference to become actionable for farmers and researchers.

One challenge lies in the “terabytes of data” camera images contain, Amerige says. “If you’re in the field, you’re dealing with Wi-Fi, or maybe satellite communication — that’s a thin pipe to the internet. So somehow the stages in the data pipeline, from edge capture through processing and presentation, need novel improvements to enhance usability under real-world bandwidth constraints to deliver an intuitive, decision-focused experience for farmers and researchers.”

By making this digital information quickly available and easy to use, farmers can better predict things like which varieties will thrive, the best times to plant, when fields will mature, and which pests or diseases may strike. These predictions guide their decisions about purchasing seeds, fertilizers, logistics and more.

Coding Connections

In the past, farmers made educated guesses to prepare for the season. “People had to rely upon their years and decades, or even family generations, of experience,” Amerige says. 

Creating technology that can distill agricultural data into something more accessible, he says, allows people who don’t have that personal or generational knowledge to succeed in farming from the start.

This meshes with one of Amerige’s personal missions: helping people by making data usable and actionable. To this end, he also serves as an Extension Master Gardener volunteer, where he shares research-based gardening knowledge with the public through talks at libraries and works as an administrator on the organization’s online Plant Toolbox. He also writes for Triangle Gardener, where he offers suggestions on gardening techniques and choosing different plant varieties. Outside of his university and Extension work, Amerige is developing an app through GitHub that will recommend plants and soil modifications based on a homeowner’s soil test results.

Reberg-Horton sees Amerige’s expertise in software and programming development as a boon for the Department of Crop and Soil Sciences’ efforts to expand the capabilities of precision agriculture. He believes AI’s use in mapping and decision support for farmers will increase in the next 20 years, starting with high-value crops and gradually transitioning to main agronomic crops. 

“Software development is at the heart of all of this, so we need developers like Steve alongside agricultural scientists to contribute to this future,” Reberg-Horton says. “His industry experience alone will help us network with the ag tech companies that are developing here in North Carolina.”