top of page


I devoted large part of my carreer and R&D work in the development of AI algorithms and systems inspired by neural networks that aim to automate and optimise business processes. The vision of creating autonomic systems which exhibit various degrees of independence to achieve their goals has been central to my work in mobile robotics and agent-based process management. To achieve this vision, I had to develop novel AI architectures based on multi-agent systems and combine symbolic AI with connectionist approaches. Hopefully, the work guides others towards the advancement of AI and Autonomic Systems.

Optimisation and Reinforcement Learning

As part of my PhD studies and follow up research work, I developed the Guided Local Search algorithm. Inspired by re-inforcement learning and neural networks, the technique is currently considered one of the best meta-heuristic algorithms for the classic Travelling Salesman Problem and vehicle routing applications.

Optimisation and RL resources

Programming Console

Neuro-Symbolic AI

I am interested in approaches that bridge traditional symbolic AI with probabilistic and connectionist approaches. In this context, I developed Hierarchical Behavioural Control systems for mobile robots using the concept of Fuzzy Decision Trees. Rule-based systems for automated hardware design and neuro-fuzzy hybrid systems used in multimodal interface were amongst other projects I worked on.

Neuro-Symbolic AI resources

Robot in Nature

Autonomic Business

Large part of my R&D work has been devoted to the application of AI to automate and optimise business processes. The vision of autonomous agents managing business processes was explored in early projects. The practical applications of AI and optimisation in service-centric organisations with focus on Telecoms were presented in my book on Service Chain ManagementDuring my time at Gartner, I fostered the company's work on process mining, robotic process automation and autonomic systems.

Autonomic Business resources

Data Processing
bottom of page