ALUMNUS JOHN KOSTER: From big science to big industry

 

Karen Greene
for Illinois Physics Condensate

John Koster knows all about big. From his doctoral training at Illinois Physics to his current role as principal engineer at Amazon, collaborating on large-scale projects has been a constant.

In his graduate student days, Koster did research at the intersection of nuclear physics and high-energy physics. He was part of a 500-person international collaboration called the Pioneering High Energy Nuclear Interaction eXperiment, or PHENIX, the largest of four experiments using data from the Relativistic Heavy Ion Collider (RHIC) at Brookhaven National Laboratory.

“I am genuinely happy at Amazon. It’s a growing group, and there are a lot of challenges that I find exciting to work on, especially in the machine learning space.” 

RHIC is the only operating particle collider in the U.S., and physicists regularly send beams of heavy ions or polarized protons on a collision course along its 2.4-mile accelerator ring to study the state of matter that existed just after the Big Bang and to understand the spin substructure of protons.

The graduate student experience at Illinois Physics was a big opportunity for Koster, who had come from a small liberal arts school in Maine. Koster remembers the one-on-one mentoring he received and the collegial atmosphere of the department and says these were just as important to his education as the big opportunities or big research projects.

 The PHENIX experiment detector, called the Muon Piston Calorimeter, is installed at Brookhaven National Lab in 2006. Illinois Physics alumnus John Koster built the detector at Loomis Lab while a student, then drove it to Brookhaven National Lab to install it. Photo courtesy of John Koster

“It was good for the physics but also for the comradery,” he recalls. “We were not disappearing into a library. The Nuclear Physics Group emphasized group learning, so there was a practical aspect but also a social aspect.”

As one of Perdekamp’s students, Koster had the chance to build several different particle detectors designed to measure ionizing radiation and high-energy photons. Those projects required teamwork and cooperation, and Koster participated in everything from assembling the hardware and calibrating the instruments to performing measurements and collecting data.

At Loomis Lab in Urbana, Koster was on the team that built a detector called the Muon Piston Calorimeter, and Koster drove it to Brookhaven in Upton, NY, where it was integrated into the PHENIX project. Koster stayed on to operate the instrument. Then in May 2006, he moved from Illinois to Long Island to work full-time on the electromagnetic calorimeter, measuring the energy and spin of high-energy photons produced by collisions of polarized protons at the RHIC.

Koster’s work at Brookhaven focused on understanding the spin of protons. The direction and strength of a proton’s spin determines its magnetic and electrical properties. By smashing protons together at speeds near the speed of light, researchers seek to understand the properties of spin, including direction, momentum, and energy, and to determine how specific particles within protons, such as gluons and quarks, contribute to spin.

Koster and the Perdekamp research team explored the spin substructure of protons by alternating the colliding protons’ spins and measuring asymmetries in the results of the collisions. The work required collecting hundreds of terabytes of data, using efficient algorithms to parse and interpret datasets, and performing statistical analyses on the results.

Although it might seem esoteric to those without a physics background, understanding spin does have practical implications. For example, magnetic resonance imaging (MRI) depends on aligning the spin of protons in hydrogen atoms in the body with the strong magnetic field of the MRI.

Koster completed his PhD in 2010. His thesis on the properties of spin using the Muon Piston Calorimeter earned an award for outstanding thesis research. He then took a postdoctoral position at RIKEN, a Japanese scientific research institute just outside of Tokyo, but remained stationed at Brookhaven to continue his work on the PHENIX experiment.

In 2013 Koster decided to leave the research sector for business. Once again, he went big, joining Amazon, initially as a software development engineer.

“I was excited about all the other opportunities out there, and the things I was going to be doing at Amazon seemed interesting and fit well with my experience,” he says.

Through the PHENIX project Koster had learned to work with big data, and Amazon offers plenty of data. Now as a principal engineer, Koster works with marketing teams to analyze the performance of advertising products and to present aggregated results to Amazon’s advertisers. It’s a job that requires skills in big data and machine learning, as well as the ability to ensure data quality, speed, and reliability.

“I am genuinely happy at Amazon,” says Koster, who lives in the Seattle area with his wife, Qin Liu (the couple met while working at Brookhaven), and their two young sons. “It’s a growing group, and there are a lot of challenges that I find exciting to work on, especially in the machine learning space.”

He credits his education from Illinois Physics—and particularly the advice and guidance he received from Perdekamp—with helping him develop the scientific and social skills to succeed in both research and business settings.

Being part of the international PHENIX collaboration taught Koster how to “influence without authority.” The communication skills needed to succeed in an international collaboration—working with experts in different fields and from different countries, using the terms and language they understand—helped Koster to succeed at Amazon and to navigate the challenges of the COVID-19 pandemic.

“There are not many traditional bosses in a 500-person collaboration,” he explains.  “To influence the collaboration’s direction, you need to be able to communicate effectively, use data convincingly, and present clearly.”

The RHC complex has a lot of moving pieces, Koster adds, and whether or not their hardware is ready, scientists must follow the collider’s fixed schedule, which typically seals up in January and opens again in June.

“Working backwards from a date is extremely important. If you push to add more features to the project, you might miss out and need to wait a year until the collider operates again,” he adds.

His doctoral training and the PHENIX experiment also helped Koster to understand the importance of the science of data, including how to handle it, clean it, ensure its provenance, and protect it from biases. But beyond Illinois Physic’s depth and breadth of expertise and the chance to work with renowned scientists, Koster says the most important skill he learned as a physics graduate student was problem-solving.

As an engineer who now works to build tools that are scalable and can be used for years to help advertisers with their bottom lines, a strategy for problem solving is essential.

“You can have all the knowledge in the world, but translating that into action is what’s important for success. That’s one thing my time at the University of Illinois taught me. Learn from what others have done before you, create a strategy, break the problem into pieces, and collaborate with your team members to drive forward.”