Computer Science, Data, and AI in Pedagogy and Research.

01AI in education

How does the learning environment change when students have access to generative tools?

I am not interested in treating AI as an isolated novelty or gimmick. I am far more drawn to the question of how learning environments transform when students have continuous access to generative tools: when AI helps grasp a difficult concept, when it erodes cognitive autonomy, how to structure assignments, how to evaluate the learning process rather than only the final output, and how to verify knowledge transfer without tool assistance.

I organize research into a coherent pipeline: literature → claims → constructs → research questions → didactic tasks → empirical data → conclusions → publications. An example trajectory: cognitive offloading → AI in debugging → generator vs. reviewer → transfer test without AI → empirical result → publication.

02Machine learning and missing data

Missing data and predictive model quality.

A second research direction involves handling missing data in logistics datasets and applying machine learning models. This work grew out of my master's thesis and a conference presentation; it currently encompasses a monograph chapter on container freight rate prediction in maritime transport — including a missingness audit, MCAR scenarios, comparing deletion strategies with imputation techniques, and evaluating result stability across repeated runs.

This represents a classic academic research path: data pipelines, statistical and ML methods, rigorous validation, publications, and potential continuation in doctoral research.

03MCDA and decision pedagogy

A ranking is not a decision.

I treat the Academic MCDA Calculator as a didactic and analytical tool. It is well suited for coursework, laboratory sessions, and working with students to demonstrate that any decision analysis outcome depends directly on explicit assumptions: criteria selection, weights, normalizations, methods, and preference thresholds.

04Conceptual work

Sandbox — long-term conceptual work.

Sandbox is a working concept of an environment where agents do not possess automatic, omniscient access to the world state. Knowledge is formed locally: through observation, event traces, memory retention and decay, social ties, needs, and situated decisions.

At this stage, I do not treat Sandbox as a commercial product or a game. It is a long-term conceptual and exploratory research initiative for testing ideas around social simulation, autonomous agents, memory architectures, distributed local knowledge, and emergent behavior derived from rules rather than hardcoded scripts.

world fact → event → trace → observation → assertion → local knowledge → memory → interpretation → decision → effect

Status: conceptual work and prototype models. Not a public product.