CV

Education

• 2015–2019 PhD in Mathematics at the University of Ljubljana, Slovenia
  Thesis: Evolutionary Dynamics on Evolving Graphs

• 2011–2015 Master of Mathematics at the University of Ljubljana, Slovenia
  Thesis: Evolutionary Dynamics, Games and Graphs (awarded the Prešeren Award for students)

Certificates

• 2023 DeepLearning.AI TensorFlow Developer Specialization (issued by Coursera)
• 2022 Machine Learning with Graphs (issued by Stanford Online)
  XCS224W offered by Stanford Engineering.
• 2022 AWS Machine Learning Engineer Nanodegree (issued by Udacity)
  Awarded the AWS Machine Learning Scholarship.
• 2021 Data Scientist Professional (issued by DataCamp)

Work experience

Experimental game theory, econometric and statistical analysis of data:

• Analysed more than a 100 experimental double-auction markets with more than 80,000 individual bids and asks from both a macro and a micro perspective to gain insight into what drives the asymmetry in convergence to competitive equilibria (paper),
• Examined almost 20,000 individual actions in more than 50 experimental sessions implementing various contribution games under different feedback treatments to shed light on how individuals make decisions and learn as the game progresses (paper).

Evolutionary game theory, graph theory, programming, agent-based modelling:

• Worked on extremal problems in chemical graph theory (papers: ABC index, GG index, (external) Wiener index),
• Developed and programmed a novel clustering algorithm (papers: mPW algorithm, see also an extended version),
• Designed and implemented an agent-based model of co‐evolution of the multilayer news flow (paper).

Developed and implemented (in C#) an algorithm for drawing large networks using divisive hierarchical $k$-means, diffusion kernels on graphs, and random projections (presentation).

Research visits

Selected to take part in the Reinforcement Learning Study Group. The study group brought together organisations from industry, government, and the third sector, with highly talented, carefully selected researchers from academia to tackle a given real‐world problem by engineering data science solutions with an emphasis on reinforcement learning approaches. Specifically, I investigated how to efficiently combine graph neural networks with actor‐critic reinforcement learning to develop a robust model (refer to Section A.3.7: Graph Neural Networks on pages 26–30 of the Reinforcement Learning Study Group Report – February 2021).

Projects:

• modelling the news flow from the mass media to the population and across it,
• analysing price convergence towards competitive equilibria in experimental double-auction markets under the Walrasian mechanism, with a particular emphasis on the impact of the order-book feedback.

Projects:

• analysing the dynamics of financial transactions (the data covered nearly 3,000,000 transactions between companies and financial institutions over a decade, balance sheets, bankruptcy information, and macroeconomic indicators),
• detecting faulty sensors and spikes in radiation levels (multivariate spatio-temporal data with battery, location, velocity, temperature, and radiation readings of more than a thousand 4,096-channel gamma-ray and neutron detection sensors).

Project:

• theoretically investigating Higman's conjecture on the number of conjugacy classes of the group $U_n(\mathbb{F}_q)$ of $n \times n$ upper triangular matrices over the field $\mathbb{F}_q$.