Hello, I’m Niki Triantafyllou

I am an EPSRC Postdoctoral Pathway Fellow at Imperial College London, developing an independent research programme in combinatorial neural algorithmic reasoning for structured decision-making.

I recently completed my PhD at Imperial, specialising in computational optimisation and machine learning, under the supervision of Dr Maria Papathanasiou in the Sargent Centre for Process Systems Engineering.

My research combines machine learning, mathematical optimisation and control to develop optimisation-compatible learning methods that are scalable, reliable and interpretable. My work includes:

  • Learning energy functions and neural representations for combinatorial optimisation with constrained inference
  • Integrating differentiable and mixed integer optimisation into learning systems
  • Developing reduced-space and multi-fidelity Bayesian optimisation methods for expensive, constrained black-box problems

I apply these methods to structured decision-making problems in personalised cell and gene therapies, energy systems and large-scale logistics.

During my PhD, I worked as an Applied Scientist Intern at Amazon on large-scale fulfilment optimisation.

Research Interests

  • Learning-Augmented and Neural Combinatorial Optimisation
  • Energy-Based Models for Constrained Inference
  • Graph Learning for Optimisation
  • Decision-Focused Learning
  • Sequential Decision-Making, Reinforcement Learning and Model Predictive Control
  • Multi-Fidelity Bayesian Optimisation