Research

My research focuses on building intelligent, autonomous, and decentralized systems that can operate efficiently in complex and dynamic environments. Over the years, this work has evolved from distributed resource management, smart grids, and energy-efficient computing toward the integration of AI foundation models, agentic AI, digital twins, and decentralized technologies.

As a research mentor, I work with master and PhD students and researchers on problems at the intersection of artificial intelligence, distributed systems, energy systems, and optimization. A central theme is understanding how AI can move beyond prediction toward systems that can reason, coordinate, optimize, and act in real-world environments.

The research is organized around three complementary directions.

AI and Foundation Model Development

We investigate how foundation models can be developed and adapted for complex engineering and IoT domains, where data is heterogeneous, temporal, distributed, and strongly constrained by physical laws. Research topics include time-series and multimodal foundation models, domain adaptation, energy forecasting, model training and evaluation. A particular focus is the development of foundation models for power grids, where models need to capture the temporal, spatial, and operational characteristics of large-scale energy systems.

An important research direction is the combination of foundation models with digital twins. We explore how AI can give digital representations of physical systems stronger capabilities for perception, reasoning, prediction, and decision support, moving toward what we describe as agentic digital twins. The goal is not simply to apply existing models to engineering problems, but to understand how domain-specific knowledge, physical constraints, data, and foundation-model capabilities can be brought together to create reliable AI for large-scale critical infrastructure.

AI Model Orchestration and Agentic AI

A second direction concerns the transition from individual AI models toward systems of models and autonomous AI agents. We study how multiple models, agents, digital twins, optimization algorithms, and software services can be orchestrated to solve complex problems that cannot be addressed effectively by a single model. This includes questions of task decomposition, model selection, coordination, planning, tool use, communication, and constrained decision-making.

A major application area is autonomous energy management. We investigate how AI agents can represent and coordinate physical assets such as generators, batteries, buildings, and other distributed energy resources. This research builds on our earlier work in distributed systems, autonomic computing and bio-inspired optimization, while introducing new capabilities enabled by foundation models and generative AI.

The longer-term objective is to develop autonomous systems that can perceive their environment, reason about goals and constraints, coordinate with other agents, and make decisions while remaining grounded in the physical systems they control.

P2P Systems and Distributed Ledgers

A third, longer-standing research direction addresses decentralized coordination and trust in distributed systems. Our work on peer-to-peer energy systems, blockchain, and distributed ledgers investigates how autonomous participants can coordinate resources, exchange energy and flexibility, and participate in local markets without depending entirely on centralized control. Research topics include P2P energy trading, local flexibility markets, virtual power plants, energy communities, smart contracts, privacy-preserving mechanisms, and decentralized optimization.

This line of research provides an important foundation for our current work on agentic AI. As AI agents increasingly become autonomous participants in complex systems, questions of trust, identity, coordination, incentives, verification, and accountability become essential. We therefore explore how distributed ledgers and smart contracts can provide the infrastructure needed for trustworthy coordination between autonomous agents and physical assets.