My current research focus is on developing quantum AI algorithms for addressing fundamental tasks related to practically-motivated problems in scientific computing. Most of my technical background has been in the quantum-inspired machine learning area of Neural Quantum States (NQS), which uses classical neural networks as trial wavefunctions for solving quantum systems using a variational Monte Carlo training algorithm. NQS represents a relatively small, but active and growing area of research within AI-for-Science, and has been demonstrated to have a broad domain of application: combinatorial optimization, high-dimensional linear algebra, physical particle simulation, quantum chemistry, and quantum state tomography. In a sense, NQS can be viewed as a fully classical analog to the Variational Quantum Eigensolver (VQE), a fundamental algorithm in near-term quantum machine learning.
I also have a broader interest in other areas of machine learning, including meta-learning, applications of deep learning toward disease modeling and prediction, computer vision, reinforcement learning, and privacy-preserving machine learning.
Peer-Reviewed Publications:
Retentive Neural Quantum States: Efficient Ansätze for ab initio Quantum Chemistry
Oliver Knitter, Dan Zhao, James Stokes, Martin Ganahl, Stefan Leichenauer, Shravan Veerapaneni
Machine Learning: Science and Technology. 2025. Available through open access here.
Toward Neural Network Simulation of Variational Quantum Algorithms
Oliver Knitter, James Stokes, Shravan Veerapaneni
Neural Information Processing Systems (NeurIPS). Workshop on AI for Science: Progress and Promises. 2022. Available here on the ArXiV.
Meta Variational Monte Carlo
Tianchen Zhao, James Stokes, Oliver Knitter, Brian Chen, Shravan Veerapaneni
Neural Information Processing Systems (NeurIPS). Workshop on Machine Learning and the Physical Sciences. 2020. Available here on the ArXiV.
My master's thesis was in the area of Diophantine approximation, more specifically in looking at regions of real n-space that are, "best," approximated by a given tuple of rational numbers. I expanded on previous work describing the geometric structure and complexity of these regions. This work was motivated primarily by the Littlewood Conjecture and may be seen either here or under, "Miscellaneous Writings," above.