profile image

Anvita Bhagavathula

Hi, I’m Anvi! I’m a third-year EECS PhD student at MIT, where I’m advised by Priya Donti. I build machine learning surrogates for physical systems that are designed with deployment and downstream decisions in mind, mainly for climate, renewable energy, and power systems applications. My goal is to make these surrogates reliable enough to stand in for expensive simulations and traditional solvers in practice.

Research

Surrogates have made a lot of progress as replacements for expensive simulation, but three recurring barriers keep them from being used in practice:

  1. Generalization: surrogates should perform reliably across operating conditions, geometries, and geographic regions without costly retraining for each new setting.
  2. Physical consistency by design: traditional numerical solvers satisfy governing equations up to a stated tolerance, whereas surrogates do not offer the same guarantees.
  3. Decision alignment: surrogates are usually trained to minimize field-level error (e.g., RMSE), which doesn’t necessarily lead to good downstream operational decisions.

I seek to address these challenges through (1) systematic benchmarking of out-of-distribution and downstream-task performance (2) inductive biases in models, such as constraint-enforcing losses and architectures inspired by numerical solvers (e.g., spectral methods) that capture multiscale structure and (3) novel training strategies that use subproblem decomposition, self-consistency, or differentiable solver gradients in the training loop, to enable models to generalize even in low-data and label-free settings.

My current projects include physics-informed neural operators that generalize to unseen length scales, with applications to accelerating computational fluid dynamics simulations, and graph-based power flow solvers that transfer across grid topologies and sizes.

About Me

I completed my undergraduate degree at Brown University, where I majored in Physics and Applied Mathematics, followed by a Master’s in Electrical and Computer Engineering at Cornell Tech. As an undergrad/master’s student, I engaged in a broad range of research experiences involving scientific computing. At Brown, I studied superconductivity in 2D graphene systems using experimental methods and Density Functional Theory (DFT) simulations in the Low-Dimensional Electronics Lab (advisors: Jia Li, Brenda Rubenstein). In addition, I developed physics-informed neural networks as part of the The Crunch Group (advisor: Somdatta Goswami). I also had the opportunity to intern at Microsoft Research (advisor: Ranveer Chandra) where I worked on creating datasets for food protein property prediction, and startups such as Aqemia (drug discovery) and Transcelestial (wireless laser communications).

Outside of work, I love singing, playing piano, composing music, and reading poetry. I’ve recently started exploring cooking and creating new recipes. I also enjoy film photography, and you can find some of my photos here!


Latest Updates