How can the skills you’re developing at K help tackle a problem you’ve seen firsthand?

For Son Le ’28, a summer research internship at Kalamazoo College meant using computer vision and deep learning to address a real-world infrastructure challenge. Here’s Son on navigating the challenges of research, working across time zones, and discovering where he hopes to take this work next:

“This summer, I worked as a research intern at Kalamazoo College under Dr. Tasnim Gharaibeh, building a computer vision system that uses deep learning to detect defects in municipal water and sewer pipelines. The project is anchored in Vietnam’s aging tropical infrastructure, a context I know personally, having grown up in Ha Tinh. My work was funded through the CCPD’s Summer Internship Stipend, which made it possible for me to commit fully to research rather than choosing between this opportunity and a paying job. 

My work involved training YOLO-based object detection models to classify pipeline defects like holes, leaks, and rust from a dataset of over 2,000 annotated images. But one of the biggest challenges had nothing to do with the model itself. I was working with multiple datasets that labeled the same damage types differently, with no universal naming convention to unify them. Even Dr. Tasnim and I found no clear answer in the literature. That constraint led us to design a two-layer cascade architecture: a simple binary classifier first, then a multi-class model on top. The limitation shaped the solution. Research does not hand you a roadmap. You study the problem deeply, brainstorm every angle, and decide with limited resources. 

Working remotely across time zones added another layer. Dr. Tasnim was in her home country juggling several projects, so our communication windows were narrow. Finding a shared time slot meant rearranging my schedule, especially since I was also completing an industry internship and preparing for the next recruiting cycle. Balancing all three was mentally the hardest thing I have done. But Dr. Tasnim’s mentorship kept me grounded. She has a way of pulling your focus back when you start scattering, and I am deeply grateful for that. 

This experience confirmed that I want to pursue software engineering with a strong research foundation. Moving forward, I hope to build on this work by proposing it as my senior thesis at Kalamazoo College. Thank you to the CCPD for making this possible and to Dr. Tasnim for trusting me with work that genuinely matters.” – Son Le ’28