South Dakota Mines Students Build Artificial Intelligence Software to Keep Critical Supplies in Stock When Needed

Everyone has felt the frustration of trying to find an important item that is out of stock. It's the job of supply chain and product managers to keep in-demand products on the shelves, but this job can be a massive challenge compounded by a huge range of variables, from basic supply and demand to overseas conflict to unstable weather patterns to labor and market disruptions to global pandemics.

Laboratory supplies are not immune to the supply chain challenges and are critical for a wide range of industries. This is a problem a team of computer science and engineering students at South Dakota Mines stepped up to solve this past year. Colton Snyder, Hunter Hufstutler and Riley Nupen spent their senior year developing artificial intelligent software for Midland Scientific Inc. (MSI) to help predict what products are needed when.

“The idea of correctly predicting the future is a common problem in business, so allowing a computer model to accurately decide when to stock a product saves on employees trying to manually look through common trends plus the model might find an unexpected trend that was not considered,” says Riley Nupen, one of the Mines computer science students on the MSI team.

MSI is a woman-owned, full-line distributor of laboratory products such as chemicals, instrumentation, general lab supplies, glassware, lab consumables, media, measurement equipment, plasticware, reagents, solutions and much more. As a full line distributor, the company represents around 500 suppliers with online offerings of over 500,000 items.  The company stocks over 4000 items that make use of AI tracking and prediction. MSI primarily serves the agricultural, educational, food, industrial and research markets with a sales force that covers the entire United States.

“We always enjoy and marvel at the intelligence the next generation of students bring into the work environment”, says Vivian Pappel Greni, Midland Scientific President and a Mines alumnae. “As with most projects, a new set of eyes often brings new approaches as well as new technical knowledge to the issues at hand.  That is what we saw with this project. The students demonstrated both new technology as well as new approaches to addressing our issues.”

The computer models developed by the student team helps inform new software that tracks items that are moving off shelves and those that aren't. This way, supply chain and inventory managers can stay one step ahead of meeting the demand for critical items.

“There were many ‘eureka moments' that we encountered while working on this project,” says Hufstutler. “Since the majority of the work I did was just trying to understand the data in order to create the best possible feature sets to give machine learning models the best odds at high performance, I was constantly updating the features after new discoveries or new breakthroughs.”

The hands-on nature of this project was a direct benefit to students. Snyder's deep dive into data science as part of the project lead him to join the Data Miners Club on campus and take part in a regional competition against other universities. “The skillset I gained by working on this project helped bring our team to achieve an honorable mention at the 2023 MinneMUDAC competition against 27 other teams from some considerably larger schools,” he says.

The project was also a win-win for the company, MSI.

“Midland Scientific works very hard to advance the fields of science and technology for the greater good of society,” says Pappel Greni. “By helping the next generation of STEM professionals, we are helping fulfill this goal. On a more microscopic level, Midland Scientific has a strong need for STEM professionals to grow our business and as such advancing their careers improves the talent pool that our company can recruit.”

The university's Department of Electrical Engineering and Computer Science includes hands-on real-world projects integrated into the curriculum, which engages students throughout their college career. Graduates with a Mines computer science degree enjoy an average starting salary of $82,998. But for department faculty, these are not the only measures of success.

“The most significant measure of outcomes for our Mines computer science students is when they succeed with opportunities like MSI provided during their capstone experience. The students took what they learned in their coursework and built a solution to save MSI money. There is no better outcome,” says Brian Butterfield, a lecturer in the Department of Electrical Engineering and Computer Science at South Dakota Mines who teaches the capstone design course.