Hey, I'm Roshith.
I'm a mechanical engineering PhD candidate at the University of Maryland, and my work is centered on thermal management of data centers and electronics. Broadly, I work at the intersection of thermal-fluid systems, heat transfer, numerical methods, scientific computing, scientific software development, and experimental thermal engineering. I am especially interested in problems where deep mechanical engineering fundamentals meet the need for system-level thinking, computational modeling, and the practical tools required to study and design next-generation thermal systems.
A major part of my PhD revolves around three core research efforts, two of which are ARPA-E COOLERCHIPS projects. In MOSTCOOL, I work on the development of system-level design and simulation tools for advanced data center cooling systems, with a strong emphasis on two-phase cooling architectures spanning chip, rack, and facility scales. This includes work on two-phase solver development, thermal system modeling, and the broader challenge of enabling physics-based design tools for next-generation data center cooling technologies. In a second ARPA-E project in collaboration with the University of Florida, I use and extend these tools to study the system-level implications of a novel membrane-assisted heat sink, connecting solver development and thermal system design to an actual emerging cooling technology.
My third project is a Northrop Grumman-supported effort on encapsulated phase change material cooling concepts. That work includes both modeling and an experimental component involving the design and ongoing characterization of a novel cold plate. This is an important part of my work because I do not want my profile to be limited to simulation/modeling alone. I care about the full engineering process of thermal systems, from understanding the governing physics, to building the tools needed to study them, to connecting those ideas back to hardware, testing, and validation.
One thing that naturally happened during my PhD is that software became a much bigger part of my work. I did not start from the identity of a software engineer. I started from thermal engineering, numerical heat transfer, and mechanical systems. But the kinds of problems I wanted to solve especially in two-phase cooling and data center thermal system design required me to build more of the tools myself. Over time, that pushed me deeper into scientific software development, solver design, and computational workflows. Today, software is not separate from my engineering work. It is one of the main ways I think, build, and solve problems.
Beyond my core research, I also spend time staying close to the current state of AI and large language models. I am genuinely fascinated by what these LLMS/tools may eventually enable in engineering from using agentic systems around thermal simulations to exploring how AI could support the design and even aid in the execution of experiments. A lot of those ideas are still evolving, but they come naturally from the combination of skills I have built through my projects.
Across all of this, what excites me most is working on thermal systems in a way that does not stay confined to one box. I like problems that demand fundamentals, modeling, tool-building, system-level thinking, and real physical intuition all at once. That is the thread connecting my work, whether I am developing a two-phase solver, studying next-generation cooling architectures for data centers, or working toward experimental characterization of new thermal hardware.