Surya Kotapati ’24


Institute for Human & Machine Cognition, Pensacola, FL

This summer, I worked at the Institute for Human & Machine Cognition (IHMC), a nonprofit research institution producing cutting-edge work in big data and machine learning, cybersecurity, and robotics. IHMC’s mission is to push the boundaries of what humans can achieve physically, cognitively, and perceptually. It intends to accomplish this mission by leveraging the interactions of humans and machines to extend our capabilities in a synergistic manner. This type of work naturally necessitates an interdisciplinary perspective, and accordingly, IHMC employs a diverse group of statisticians, computer scientists, psychologists, medical doctors, and engineers. It also follows that in its service to society, IHMC works extensively with both a wide range of big players in the private sector and with the government, and has received funding from NASA, the Army, IBM, Microsoft, Boeing, and Lockheed. The interdisciplinary approach and consequential applications, along with my desire to work on the leading edge of machine learning and artificial intelligence, meant that the prospect of contributing meaningfully to IHMC’s research was an enticing one.

My first project focused on dimensionality reduction: reducing the number of variables needed to be analyzed in a data set while retaining as much meaningful information as possible. Such transformation is necessary to reduce computational cost and simplify human interpretation when there are terabytes of data available. I was tasked with evaluating and presenting novel dimensionality reduction techniques to reduce the amount of aircraft telemetry data that engineers needed to analyze while running diagnostics. This involved extensive literature review and carefully examining source code and design documentation. My next project allowed me to get more hands-on. I worked with Hadoop and Apache Spark, software that was previously unfamiliar to me, to read in the large data set and apply the dimensionality reduction techniques I had learned earlier to select the most relevant variables for analysis. I subsequently ran unsupervised machine learning algorithms in Python to visualize the data in clusters and draw key insights. It was thrilling to supplement concepts learned in the classroom with the technical know-how gleaned from weeks on the job and to then watch the results unfold in front of my eyes.

This internship has afforded me the opportunity to expand my skill set in machine learning and big data analysis. But more importantly, it has cemented one of my primary academic interests at the intersection of statistics and computer science. This has sparked in me a desire to take more classes related to data science during the remainder of my time at Williams.

I would like to sincerely thank the Class of 1966 for their generosity, as well as Ted McPherson ’67 and the ’68 Center 
for Career Exploration for connecting me with this incredible opportunity. I am also grateful for the steady support and guidance I received from my internship supervisors, Larry Bunch and Arash Mahyari.