01
From resilient energy networks to quantum hardware and AI-guided optimization.
Developed a two-stage stochastic multi-commodity flow model for resilient energy supply under correlated chokepoint disruptions.
Elsevier · published paper
Benchmarked gate-based quantum optimization across MDKP, MIS, QAP, and MSP on IBM Heron processors under a common protocol.
IOP · published paper
Introduced an LLM-guided experimentation loop that searches adaptive solver-control policies for variational quantum optimization.
arXiv · first page
Introduced a learning-based graph shrinking approach for reducing constrained combinatorial optimization instances.
Springer · proceedings cover
Soft Actor-Critic policies tune augmented Lagrangian penalties for both classical and quantum solvers, accelerating convergence and feasibility.
Springer · proceedings cover
Employed qGANs to learn demand distributions, reducing qubit requirements with a tailored comparator.
IEEE Xplore · article page
Combined Quantum Random Access Optimisation with linear relaxation to solve large-scale procurement problems.
IEEE Xplore · article page
Diagnosed cross-size transfer for EQC reinforcement-learning policies across simulators, noisy execution, and hardware.
arXiv · first page
Converted CVRP into bounded-width quantum subproblems using Lagrangian relaxation and per-vehicle knapsack QUBOs.
arXiv · first page
Dual ascent, bundle, and augmented Lagrangian updates enforce constraints without auxiliary slack variables, reducing qubit counts.
arXiv · first page
Evaluated VQE, CVaR-VQE, QAOA variants, and compression techniques such as PCE and QRAO across NP-hard benchmarks.
arXiv · first page
Integrated decision-maker risk profiles into a quantum Monte Carlo framework for multi-supplier newsvendor problems.
arXiv · first page
Focus
Three threads run through my research. Here they are in plain words, with the work behind each one.
Delivery routes, inventory and energy supply under disruption. I model these planning problems and test whether quantum and hybrid solvers genuinely help — measuring carefully when they don’t.
9 papers
Language-model agents and learning methods that propose experiments, tune solvers and keep the evidence attached — so results can be checked, not just trusted.
4 papers · 1 project
Running algorithms on real quantum processors, building fast simulators, exploring how quantum processors will integrate with supercomputers in hybrid quantum–HPC systems, and verifying circuits end to end — the engineering that makes quantum results reproducible.
2 papers · 6 projects