1. (active) PI: Inertial neural surrogates for stable dynamical prediction, Data intensive scientific machine learning, DOE-ASCR FOA-2493.
  2. (active) PI: DeepFusion Accelerator for Fusion Energy Sciences in Disruption Mitigations, U.S. Department of Energy Fusion Energy Sciences, DOE-FES FOA-2905.
  3. (active) Co-PI: Artificial Intelligence and Machine Learning for Autonomous Optimization and Control of Accelerators and Detectors, U.S. Department of Energy Nuclear Physics, DOE-NP FOA-2785.
  4. (active) Co-PI: AI emulator assisted data assimilation, Future computing, LDRD-Prime, Argonne National Laboratory, U.S. Department of Energy. (PI - Rao Kotamarthi)
  5. (active) Senior personnel: RAPIDS2:A SciDAC Institute for Computer Science, Data, and Artificial Intelligence, U.S. Department of Energy. (PI - Rob Ross)
  6. (Finished) Co-PI: A Scalable, Energy Efficient HPC Environment for AI-Enabled Science, Collaborative PPoSS funding, National Science Foundation. (PI - Zhiling Lan)
  7. (Finished) PI: Margaret-Butler Fellowship project: Scalable machine learning for turbulence closure and reduced-order modeling (2 yr postdoctoral fellowship).
  8. (Finished) PI: SambaWF: Highly resolved surrogate models for weather forecasting, LDRD-Expedition, Argonne National Laboratory, U.S. Department of Energy.
Software development
  1. PythonFOAM, In-situ data analyses with OpenFOAM and Python.
  2. TensorFlowFOAM, A framework that enables the deployment of TensorFlow deep learning models and partial differential equation solutions concurrently in OpenFOAM.
  3. PAR-RL, A framework that leverages the Ray library to deploy scalable deep reinforcement learning for arbitrary scientific environments on leadership class machines.
  4. PyParSVD, A Parallelized, streaming, and randomized implementation of the SVD for Python using mpi4py.
  5. PySPOD, A package to compute the spectral proper orthogonal decomposition in Python.
Plain Academic