Bio
Filotas Theodosiou is a senior applied researcher and lecturer at the Research Group of Predictive AI and Digital Shift at VIVES University of Applied Sciences. He holds a Bachelor's degree in Mathematics from the Aristotle University of Thessaloniki and a Master's degree in Data Science from the University of Skovde (Sweden). He is a member of the International Institute of Forecasters (IIF), the Early Career Forecasters (ECF), and Belfors (Belgium Forecasting Society), and has contributed as a reviewer for the International Journal of Forecasting.
His work focuses on designing, modeling, and operationalizing machine learning-based forecasting solutions for complex business decisions. Beyond academic output, he has hands-on experience delivering forecasting systems in production settings: from business scoping and data pipeline design to model development, probabilistic decision support, and cloud-based MLOps deployment. His applied forecasting experience spans hierarchical systems, data-limited demand prediction, interpretable and actionable prediction frameworks, and optimization workflows for perishable products.
In parallel with his own technical delivery, he leads and supervises end-to-end student ML/MLOps projects that mirror real applied data science work. Through course and thesis supervision, he guides problem framing, solution architecture, evaluation design, deployment planning, and the trade-offs required to move from an idea to a credible product. This supervision work provides practical evidence of technical leadership and is highlighted on the Teaching page.
He also actively evaluates the evolving AI landscape and gives regular talks on practical usage, opportunities, and limitations of modern AI tools, with particular attention to how LLM-based components can be integrated into robust forecasting pipelines. In addition, he has experience with building agentic AI application, exploring multi-agent systems and autonomous workflows. Finally, he is the author of Medium articles that translate academic forecasting research into practical guidance for real-world implementation.
Research Interests
- Demand Forecasting
- Machine Learning Methods for Forecasting
- Hierarchical Forecasting
- Data Engineering for Time Series Forecasting
- Limitations and Opportunities of Foundation Models for Time Series Forecasting
Toolbox
- Excellent in problem scoping, solution architecture, technical supervision & effective communication
- In-depth understanding of time series dynamics in various applications
- Advanced Python, fluency in R
- MLOps Technologies
- Experience with cloud platforms (AWS)
- Deep understanding of LLM tools & functionalities for text/coding tasks
- Experience with agentic AI application
- Supervision of end-to-end student ML systems and production-oriented thesis work