Research vision
A recurring idea in my work is that learning need not be an isolated process. Many of the most interesting AI systems contain multiple models, agents or objectives whose interactions shape the final behaviour. Evolutionary and coevolutionary computation provide a natural framework for studying and optimizing those interactions.
My long-term research programme develops co-adaptive AI systems: learning systems in which multiple components jointly adapt under changing data, objectives, resource constraints and social or environmental requirements.
Co-Adaptive & Evolutionary AI
Evolutionary and multi-objective optimization for machine learning, with particular emphasis on coevolution, interacting populations, model cooperation and efficient search under limited supervision.
Generative & Agentic AI
Generative models and emerging agentic systems provide especially rich settings for interaction. My work studies how evolutionary mechanisms can improve robustness, diversity and adaptive behaviour when multiple learners influence one another.
Federated, Private & Distributed Learning
Distributed learning introduces heterogeneity, communication constraints and privacy requirements. I am interested in co-adaptive mechanisms that allow multiple learners to specialize while contributing to useful global behaviour.
AI for Sustainability, Climate & Cities
Methodological advances are paired with real-world problems in mobility, environmental intelligence, resource management, urban infrastructure and climate-related analysis. These domains provide demanding testbeds in which adaptivity, efficiency and responsible deployment matter.
Open research
Reproducibility, open implementations and reusable datasets are important parts of my research practice. Selected code and project materials are available through my GitHub profile.