Can we teach an AI Scientist to engineer Hamiltonians in real materials—by stacking and twisting single sheets of atoms—until a target strongly correlated phase emerges on demand? This PhD aims to integrate “intelligence” into our autonomous 2D-material fabrication pipeline with in-situ optical metrology to plan, assemble, and verify programmable quantum materials and devices, enabling fundamental discoveries about emergent quantum states. 

The project. Twisting and stacking van der Waals monolayers produces moiré superlattices with flat electronic bands and strongly correlated phases. By controlling twist angle, displacement fields, strain, and dielectric environment, we can effectively engineer the many-body Hamiltonian—tuning hopping, interaction strengths, and symmetry-breaking terms. This controllability underpins phenomena from correlated insulators to unconventional superconductivity and magnetism, but demands precise, reproducible fabrication with tight angle control and clean interfaces. The parameter space is enormous, creating a natural role for AI to automate experiments, improve material/device quality, and accelerate data analysis and interpretation. 

Training and skills you will gain: 

• Machine vision & AI: extend detection/segmentation pipelines (Python/OpenCV/PyTorch) and train decision/policy models to plan and adapt assembly steps in real time. 

• Optical spectroscopy & metrology: PL and reflectivity for material ID and interface quality; second-harmonic generation for twist/orientation; low-temperature magneto-optical spectroscopy for Hamiltonian-level readout of minibands, interactions, and symmetry breaking. 

• 2D device fabrication: autonomous pickup/release, twist/strain setting, contamination avoidance; electrical/optoelectrical validation where appropriate. 

• Cryogenics: cryostat operation, wiring, alignment etc to characterise emergent states at low temperature. 

• Quantum materials foundations: connect process variables to miniband structure and correlated phases; practice reproducible research engineering (structured logging, digital twins, versioned data/code) and engage in collaborative/industrial environments via project partners.