The future of AI goes beyond predicting the world. It will learn to model, optimize, and control it.

At the SOLARIS Lab at Johns Hopkins University, we are pushing the frontiers of Scientific Machine Learning (SciML), bringing together machine learning, applied mathematics, optimization, and control theory to develop a new generation of intelligent systems that understand physical laws, make optimal decisions, and operate reliably in the real world.

Our research explores fundamental questions at the intersection of AI and scientific computing:

  • Learning to Model (L2M): Physics-informed learning, neural differential equations, and neural operators for complex dynamical systems.
  • Learning to Optimize (L2O): Learning algorithms that accelerate or replace expensive numerical optimization, from continuous to combinatorial problems.
  • Learning to Control (L2C): Differentiable predictive control, neural operator policies, and autonomous decision-making for complex physical systems.
  • Next-Generation Scientific AI: Composable differentiable programs, learning in infinite-dimensional function spaces, and scalable AI with mathematical guarantees on safety, feasibility, and real-time execution.

We are looking for exceptional thinkers and builders with backgrounds in Applied Mathematics, Computer Science, Engineering, or related fields. Ideal candidates combine rigorous mathematical foundations with strong programming skills (particularly Python, PyTorch, or JAX), intellectual curiosity, and the ambition to tackle fundamental problems rather than incremental improvements.

If you are passionate about developing new mathematical ideas, turning them into working algorithms, and pushing the boundaries of what AI can achieve, we want to hear from you.

Join Us

PhD applicants: Apply through the Johns Hopkins graduate admissions portal. See PhD application guidelines for further information.

Prospective researchers: Send your CV, research interests, and relevant research or coding projects to jdrgona1@jh.edu.

Explore our SciML tutorial at L4DC 2026 to get a taste of the research directions we are pursuing.