I work at the intersection of machine learning, optimisation, and systems engineering, with a strong bias toward clarity, robustness, and real-world constraints. My interests tend to cut across domains: from mathematical modelling and optimisation, to energy systems and infrastructure, to the design of efficient, interpretable ML pipelines. What ties these together is a focus on structure — understanding how complex systems behave, where their leverage points are, and how theory translates into practice. I am less interested in isolated techniques than in the architectures that make them useful, maintainable, and resilient over time.

I am exploring how machine-learning classification models can be used to reduce the energy consumption of retrieval-augmented generation (RAG) systems, by activating only the components that are necessary for a given query.

The projects and notes you will eventually find here are small slices of that work: research logs, infrastructure sketches and experiments in making ML systems more aligned with physical and organisational constraints.