Goucher senior studies solar eruptions at NASA internship

Last summer, Topper brought the experience he gained in Goucher classrooms to NASA’s Goddard Space Flight Center. Through an internship in their Heliophysics Science Division, he worked on developing a machine-learning neural network to predict solar wind structures and their magnetic field orientation—a project that could help NASA predict and avoid dangerous space-weather effects.

When Topper—a computer science and integrative data analytics double major—began his search for a summer internship, he was unsure of where to start. He reached out to the professors in his department, who quickly provided him with a list of opportunities in the area. Among the suggestions, the Goddard Space Flight Center jumped out as an attractive option.

After applying, Topper connected with his professor Thomas Narock, who happened to know someone in the heliophysics division.

According to Narock, it can be challenging for students to see the connection between course material and opportunities outside of college.

“I believe the role of Goucher faculty is to help students recognize these connections,” Narock said. “In Jack’s particular case, a discussion after class on a topic he was interested in led to an independent study project and ultimately to a summer at NASA.”

Topper said Narock’s Advanced Machine Learning class prepared him for his work at NASA.

“There are elements of the internship that threw me in the deep end,” Topper said. “But most of the work coincided with my machine learning class. If I hadn’t taken that class, I don’t think I would have even gotten the position.”

During the internship, Topper used a form of artificial intelligence trained on data sets provided by NASA to improve their ability to predict flux ropes—twisted magnetic field lines caused by the sun’s eruptions—observed by NASA’s Wind spacecraft.

In December, Topper gave a presentation at the American Geophysical Union on the impacts of the neural network he used. He found that changes to the network significantly improved predictions on flux ropes observed by Wind. Out of 167 solar wind eruptions, the neural network reduced 49 incorrect predictions to only 10.

“The internship and presentation were fantastic experiences,” Topper said. “It taught me a lot of lessons in professionalism and understanding what the workforce actually looks like. Getting to go into the Goddard center and see firsthand what people are working on was a huge learning experience.”

Topper graduates this spring and is currently looking for positions at NASA or for other machine learning roles. He was recruited to Goucher to play as a part of the school's Men's Soccer team. He said he was drawn to the school due to the its small class sizes, and it's that one-on-one interaction with professors that made his career experience great.

“It was knowing that people were going to know me and take care of me that brought me to Goucher,” Topper said. “And that’s exactly what happened. When I was struggling with questions in my internship, they were helping me fix my problems and guide me through the experience.”