Pushing the Boundaries of Computational Fluid Dynamics - Profile Dhruv Apte

Dhruv Apte’s research journey spans continents, fluid phases, and climate systems. A Peter O’Donnell, Jr. Postdoctoral Fellow at the Oden Institute for Computational Engineering and Sciences, Apte brings a multidisciplinary background in aerospace engineering, scientific machine learning, and earth system modeling.

He completed his Ph.D. at Virginia Tech, where his research focused on cloud cavitating flows, which are complex vapor-liquid interactions that occur in turbulent environments. By integrating machine learning into traditional turbulence models, Apte helped improve the simulation accuracy of these highly unsteady, multi-phase flows. His research included collaborations with the University of Melbourne in Australia and the Institut Pprime in France. 

“That work focused on fractions of a second,” he explained. “But now I’m simulating 30 to 40 years into the future. The scale has changed dramatically.”

Today, Apte is part of the Computational Research in Ice and Ocean Systems Group at the Oden Institute at The University of Texas at Austin, where he works with Professor Patrick Heimbach on improving modeling of ocean-ice interactions as part of a larger multiphysics Earth-system model. His current research explores how neural operators and physics-informed machine learning can reduce the computational costs of simulating climate systems. “We already have satellite data, and we understand the governing physics,” he said. “But traditional models still require massive computing power. The question is, can we build faster, trustworthy surrogates that honor both the data and the physics?”

One of Apte’s projects applies neural operators to act as surrogates to expensive general circulation models by learning from historical satellite records. He is also contributing to the development of surrogate modeling frameworks in the Julia programming language to enable faster, scalable Earth system models by implementing automatic differentiation, multi-level intermediate representations (MLIR) and checkpointing schemes with colleagues Joe Kump, Sarah Williamson and other collaborations across various institutions.

Apte praised Heimbach’s mentorship for fostering technical depth, scientific independence and genuine personal support. “Patrick helps clear the runway so that I can take off,” he said. Heimbach was also the first recipient of the CSEM Advising Award, recognizing his outstanding dedication to promoting student success. 

In addition to his technical goals, Apte is thinking about the broader environmental and professional impact of his work. He wants to build machine learning tools that are not just accurate, but also energy-efficient, to reduce the carbon footprint of large-scale simulations. “If we can replace a full large-scale model with a fast surrogate, we make science more sustainable and accessible,” he explained.

He also values the opportunity to mentor students and refine his own skills as a communicator and collaborator. “Being a postdoc is a unique window,” he said. “You’re still close to the technical work, but you start thinking about leadership and long-term vision too.”

Outside of research, Apte enjoys dancing, swimming, and cooking. “When I’m stressed, I cook something slow and methodical in the kitchen or head to the campus gym to swim. It helps me reset,” he said. Since moving to Austin, he has enjoyed exploring the city’s energy and outdoor spaces. “There’s so much happening here. Whether it’s going to dancing socials or catching live music, Austin has been a great place to recharge.”

When asked what motivates him, Apte doesn’t hesitate. “Ocean science affects everyone. Whether it’s sea level rise or storm forecasting, the stakes are huge. Knowing that my work could help us better understand and prepare for the future keeps me going.”