Stochastic and Mathematical Optimization
Developing mathematical programming and decomposition-based approaches for operational decision-making under uncertainty.
Research
My research sits at the boundary of operations research and real-world decision problems — resilient supply chains under uncertainty, e-commerce and last-mile operations, AI-enabled service innovation, and graph-based learning for biomedical data.
Research philosophy
Understanding uncertainty.
Every model is an attempt to understand uncertainty — not to eliminate it.
Core Research Pillars
Supply Chain Resilience and Analytics
Studying resilient network design, disruptions, routing, inventory systems, and data-driven supply-chain decisions.
Game Theory and AI-enabled Service Innovation
Analysing strategic interactions, equilibrium behaviour, and AI-driven service technologies in supply-chain and e-commerce systems.
Graph and Hypergraph Learning
Exploring graph-based and hypergraph neural-network methods for modelling higher-order relationships in biomedical data.
Methodological toolkit
The methods and tools I draw on across projects. These reflect confirmed working methods rather than proficiency levels.
Mixed-Integer Linear ProgrammingTwo-Stage Stochastic ProgrammingSample Average ApproximationLatin Hypercube SamplingClassical and Accelerated Benders DecompositionGame-Theoretic ModellingEquilibrium AnalysisRoute OptimizationCausal InferenceAugmented Inverse Probability WeightingMachine LearningDeep LearningGraph and Hypergraph Neural NetworksPythonIBM ILOG CPLEXGurobiWolfram MathematicaPyTorchTensorFlow
Application areas
- Resilient FMCG supply chains
- E-commerce supply chains
- Last-mile delivery
- Supply-chain disruptions
- AI-enabled service innovation
- Delivery-risk analytics
- Biomedical data modelling
Approach
I am drawn to problems where the gap between theory and operational reality is widest. Optimization models are only as useful as the assumptions behind them — which means understanding the domain, the constraints, and the people making decisions, before reaching for the solver.
My methodological instinct is to start with the structure of the problem: what decisions need to be made, at what point in time, and under what uncertainty. From that structure, the model often becomes apparent. The goal is always a formulation that is rigorous enough to be solved reliably and interpretable enough to be trusted in practice.
Current questions
- How can decomposition methods such as accelerated Benders scale resilient multi-echelon supply-chain design under disruption uncertainty?
- How do AI-driven service technologies, such as chatbots, reshape strategic interactions and equilibrium outcomes in e-commerce supply chains?
- Can hypergraph neural networks capture higher-order relationships in biomedical data more faithfully than pairwise graph models?
If you work on problems at the intersection of optimisation, supply-chain resilience, game theory, or graph-based learning — or if you are exploring how data-driven models can support better decisions — I would be glad to connect.