Scientific Machine Learning and Operations (STARS) at the Department of Computational and Data Sciences is under the Division of Interdisciplinary Research at Indian Institute of Science Bangalore, India.

We develops accurate, efficient and robust parallel numerical (finite element) schemes for solving partial differential equations (PDEs),  Scientific Machine Learning and High Performance Computing.

Please visit our in-house finite element package ParMooN.

STARS (Scientific Machine Learning and Operations)

STARS (Scientific Machine Learning and Operations) is a research group at the forefront of scientific computing, machine learning, and high-performance computing. Our primary focus revolves around two pivotal areas: Scientific Machine Learning (SciML) and Machine Learning Operations (MLOps). Within SciML, we are pioneers in developing innovative machine learning algorithms to solve complex closure problems, harnessing the power of Physics-informed neural networks (PINNs) to augment numerical schemes, and utilizing AI-based techniques to estimate uncertain model parameters. Additionally, we leverage data-driven approaches to revolutionize traditional scientific simulations, enhancing their accuracy and efficiency.

In parallel, our emphasis on MLOps drives us to push the boundaries of scalable ML algorithms and distributed training techniques. We delve into the realm of cloud computing to enable seamless deployment, ensuring ML models can handle large-scale datasets effortlessly. With an unwavering commitment to robustness and efficiency, we establish ML model and data version control, design and implement CI/CD pipelines, and streamline the entire ML deployment and operations process. By placing a strong focus on MLOps, we empower organizations to effectively integrate machine learning into their workflows, facilitating real-world applications and tangible impact.

While SciML and MLOps form the crux of our research, we also excel in other areas such as Data Science, where we pioneer anomaly detection, predictive behavior modeling, and personalized recommender systems with a focus on fintech applications. Additionally, our expertise in Computational Science extends to finite element analysis, multiphase flows, fluid-structure interactions, turbulent flows, and epidemiological events, employing hybrid CPU-GPU parallel algorithms and hardware-aware scalable parallel implementations.

In summary, STARS is a leading research group dedicated to advancing scientific computing, machine learning, and high-performance computing. Our unwavering focus on Scientific Machine Learning (SciML) and Machine Learning Operations (MLOps) positions us at the forefront of cutting-edge research, driving innovation and delivering practical solutions to complex problems in academia and industry.



  • AI and MLOps for Defence


Learn more


Finite Element Methods

  • Finite element methods for the solution of PDEs and surface PDEs
  • Multigrid methods, ALE approach for moving meshes
  • Variational multiscale methods based Turbulence modelling for fluid flows
  • Higher order discontinuous Galerkin schemes for hyperbolic PDE’s
  • Fluid-structure interaction problems, ship hydrodynamics
  • Turbulent flows with moving/deforming solid bodies

Learn more


High Performance Computing

  • ParMooN, an open-source finite element software development
  • GPU-Accelerated Algebraic Multigrid solvers


Learn more



Scientific ML

  • Computational models for particle deposition in human air pathway
  • Biophysical model of cancer invasion
  • ML based stabilisation schemes for Singularly perturbed PDE’s
  • Physics Informed Neural Networks for Fluid flows
  • Reduced order modelling for fluid flows


Learn more