Postdoctoral Research Associate in Control Systems, Artificial Intelligence, and Scientific Machine Learning for Fusion Energy

Lehigh University invites applications for a Postdoctoral Research Associate to advance control systems engineering, artificial intelligence, and scientific machine learning for nuclear fusion energy. The role focuses on developing control algorithms, neural surrogate models, state estimators, scenario optimization workflows, actuator management strategies, and digital-twin simulations for tokamak plasmas, along with publishing research and mentoring students.

What you'll do:

• Neural Surrogate Modeling: Develop, train, and validate fast neural-network surrogate models (e.g., transport surrogates, edge surrogates, free-boundary MHD surrogates) for real-time predictions and control-oriented execution
• Advanced Control Synthesis: Synthesize and computationally test model-based (Model Predictive Control - MPC), data-driven (Reinforcement Learning - RL), and hybrid (RL-MPC) multi-input multi-output (MIMO) controllers for kinetic, profile, equilibrium, divertor detachment, and burn regulation
• State Estimation & Observers: Design and implement state estimators, Extended Kalman Filters (EKF), neural observers, and physics-informed "virtual sensors" for real-time plasma state estimation and boundary/equilibrium reconstruction from limited, noisy diagnostic measurements
• Scenario Optimization: Develop plasma scenario optimization workflows leveraging nonlinear programming, genetic algorithms, and reinforcement learning for ramp-up, full-discharge, ramp-down, and burning-plasma-transition trajectory generation
• Actuator Management & Arbitration: Formulate multi-objective actuator management, reference governors, and arbitration strategies to coordinate competing actuators, prevent proximity to instabilities (e.g., NTMs), and ensure machine protection
• Digital Twin Integration: Integrate neural surrogates, plasma transport solvers, and closed-loop control algorithms into MATLAB/Simulink end-to-end predictive workflows based on LU-PCG’s COTSIM (Control Oriented Tokamak SIMulator) for in silico closed-loop validation
• Publication & Dissemination: Prepare and publish research findings in top-tier peer-reviewed scientific journals and present results at national and international control, AI, and fusion conferences
• Mentorship & Service: Assist in mentoring graduate and undergraduate students in control theory, machine learning, data analysis, and software implementation; assist in grant proposal development

What they're looking for:

• Doctoral degree in Control Engineering, Electrical Engineering, Applied Mathematics, Computer Science, Data Science, Mechanical Engineering, Physics, or a closely related quantitative field, completed by the start of the appointment
• Strong theoretical and practical background in Control Systems Theory (e.g., MIMO control, state-space methods, optimal control, model predictive control, system identification) AND/OR Artificial Intelligence / Machine Learning / Data Science (e.g., deep neural networks, surrogate modeling, reinforcement learning, scientific AI)
• Willingness to work on interdisciplinary problems at the intersection of AI, control, and physical sciences
• Demonstrated ability to conduct original research with a strong track record of publications in peer-reviewed scientific journals or premier conference proceedings
• Proficiency in scientific computing languages and environments such as MATLAB/Simulink, Python (PyTorch, TensorFlow, SciPy), or C/C++
• Strong writing, verbal, and interpersonal communication skills
• Commitment to fostering an inclusive research and teaching environment
• Proven ability to work independently and as part of a collaborative, multidisciplinary, and multi-institutional research team
• Prior research experience in plasma control, tokamak magnetic confinement fusion, or computational fusion science
• Artificial Intelligence (AI), Machine Learning (ML), and Scientific Machine Learning (SciML) applied to physical or engineered systems, including deep neural network surrogate modeling, physics-informed neural networks (PINNs), reinforcement learning (RL), transfer learning, dynamic system surrogates, or uncertainty quantification
• Model Predictive Control (MPC), data-driven control, or hybrid model-based / data-driven controller synthesis (e.g., RL-MPC) for complex dynamical systems
• State estimation, Extended Kalman Filters (EKF), neural observers, physics-informed virtual sensors, or real-time diagnostic mapping

Benefits:

• Benefits
• Comprehensive healthcare plans
• Tuition remission
• Retirement savings opportunities
• Professional growth
• Personal development
• A strong sense of community