Research

I’m currently conducting research on Autonomous Programming under the supervision of Dr. Shreyas Sundaram, a Marie Gordon Professor at the Elmore Family School of Electrical and Computer Engineering at Purdue. Dr. Sundaram is an expert on control systems and large-scale networks. My research with Dr. Sundaram focuses on exploring how algorithmic logic meets self-evolving systems—essentially looking at how code can adapt autonomously. After spending three years as the lead autonomous programmer for my VEX Robotics teams in grades 7–9, I decided to dive deeper into the mathematical techniques and models behind autonomous programming.

Fall 2025 – Spring 2026 Research

During this time, my research project focused on developing various algorithms to speed up the lap times of quadcopters through gated racing courses in the high fidelity simulation platform AirSim. These algorithms, which ranged from spline trajectories all the way to curvature-based velocity modulation, all were autonomous; the drone was being controlled solely by the algorithms and there was no user interaction with the drone. After testing on various racing courses and analyzing the data, I found that the tuned curvature-based velocity modulation algorithm, which approximates the local curvature of the drone’s path at all times, and autonomously adjusts velocity based on this computed curvature, was able to reduce the lap time the most.

Summer 2026 Research

During this time, with guidance from Prof. Sundaram and my graduate student mentor Josh, my research focus shifted to multi-agent navigation and exploration of indoor environments. Specifically, I worked on getting a team of four drones to autonomously and cooperatively navigate unknown spaces in the form of mazes. I began approaching this research question at a single-agent scale similar to my Fall 2025 to Spring 2026 research. I coded a custom maze environment in AirSim and implemented an A* algorithm to find the shortest path between two cells in the maze. Next, I created eight different drone path-tracking controllers and tested them in a full-factorial sweep of controller, flight mode, target speed, and lookahead distance — 220 valid treatments across 660 total runs — in order to find the fastest way to have a single drone navigate from the start cell to the end cell of a maze. I found that a variable-speed pure-pursuit controller (similar to the curvature-based velocity modulation from my previous research) yielded the fastest lap time at 15.98 seconds, roughly 23% faster than AirSim’s built-in path follower.

After establishing this single-agent baseline, I scaled the problem up to a team of four drones exploring a maze whose layout was entirely unknown to them at the start. Rather than relying on a central planner to assign routes, I implemented a decentralized coordination policy built from four local rules: wall following to advance into unexplored cells, frontier detection to track which cells remained unvisited, a task-allocation rule that dynamically assigns frontier cells to drones based on distance and estimated arrival time, and an obstacle-avoidance rule in which drones yield right-of-way at perpendicular junctions. Each drone used the variable-speed pure-pursuit controller from the single-agent method to fly to assigned cells, which connected the two halves of the project. I tested the system across maze sizes from 6×6 to 12×12 and across three levels of maze complexity, and found that the team reached full coverage in every configuration tested. In addition, the system was robust across mazes of different complexity (both very connected mazes with few dead ends and very disconnected mazes with many dead ends). Completion time scaled approximately linearly as maze size increased. I also found that the exploration workload distributed itself evenly across the four drones from the task-allocation rule alone, without any explicit load-balancing logic.

I presented this research in the form of a research talk at the Summer 2026 Purdue Undergraduate Research Symposium and was awarded high distinction among over 500 research presentations.

Here is my abstract for the research symposium:

Here is one of my practice run-throughs of my research talk:

STEM Teaching

INSPIRE/SCALE K-12 Summer Camps

I have been a part of the INSPIRE/SCALE K-12 program since 8th grade. Every summer, I work alongside an incredible team of graduate students, undergraduates, and professors from Purdue to provide STEM activities for students from grades 1-9. This successful program, led by Dr. Morgan Hynes, has been running for many years, and I’m proud to contribute to its mission.

My role includes assisting in developing new STEM activities, teaching STEM concepts, and collecting research data on how students learn these targeted STEM concepts. I love the creative process – designing new activities, piloting them with the team before the camps, and interviewing campers to hear about their learning experiences and new knowledge!

INSPIRE camps aim to spark interest in microelectronics and encourage students to explore careers in this exciting field. Over the years, we’ve taught a wide range of activities – from coding with micro:bits to building circuits on breadboards. Check out this website to learn more about Micro:bits! https://microbit.org/

It is an honor to collaborate with such a dedicated team and make an impact on hundreds of K-12 students from culturally, ethnically, and socioeconomically diverse backgrounds. I am excited to continue being part of this amazing program!

Check out our camps here, https://engineering.purdue.edu/INSPIRE/Workshops/summer-camps and here, https://engineering.purdue.edu/INSPIRE/Resources/scale_k-12

DesignSTEM Conference for Students

I was thrilled to join the DesignSTEM team in 2025! The CATALYST team at Purdue, led by Dr. William Walker, has been offering virtual STEM “conferences” to thousands of elementary and middle school students from rural-area schools since 2018. Each fall and spring, more than 600 elementary students participate in the program.

My role at DesignSTEM includes designing new STEM activities for the program, assisting with teaching students in Zoom sessions, and organizing materials to be shipped to participating schools. For the May 2026 conference, I created an activity that combines electricity, engineering, and coding using Micro:bits, an exciting way to bring hands-on learning to students virtually! Check out my poster presentation below!