Data Scientist • Developer • Researcher
Mihaly Hanics
Data scientist, developer, and researcher from Hungary, active in Vienna & Budapest. Especially interested in
graph data solutions, AI/ML and complex systems.
My interests include, but are not limited to:
• AI agents: I'm working on docx-llm, a structured document generation product using AI agents.
• Graph (data) solutions: I went from graph theory / operations research to network science, analytics, Neo4j and knowledge graphs. Now: GNNs and graph processing at scale.
• Complex systems: systems with many emergent features. My MSc thesis gave a method based on causal inference to analyze such effects in any dynamical system. Relevant for DL/AI.
• Art: E.g. the
PainterPalette dataset
I created, the largest historical dataset of painters, with 10000+ artists and 29 attributes. Presented at NetSci 2025.
I am also reading on data engineering, biomedical data science, Neural Algorithmic Reasoning, Neural CA, and hope to start creating content soon!
NetSci 2025 (June 2025): Gave a talk in Maastricht on painter networks and the
PainterPalette dataset
at the Understudied Networks Symposium!
Delightful for the opportunity!
Recently finished my MSc in Data Science at CEU, Vienna, Austria where I've worked on 3 major projects:
• Public road network construction, commuters estimation: From geospatial data of roads and traffic data, I constructed a graph of Hungary's daily public road traffic, and implemented graph-based methods to estimate the origin-destination matrix (town-town commuters). Supervised by
Marton Karsai
and
Gergely Odor.
• PainterPalette and art projects: Contributed in various ways to quantitative research in art. Built the PainterPalette dataset, created a pipeline to scrape contemporary artists and connect them based on exhibitions, a SQL database of paintings, artists, styles and institutions, and many analyses.
• Causal framework for complex system properties: Developed a novel framework to measure synergy, redundancy, etc. in dynamical complex systems, based on local causality instead of information theory. Tested using cellular automata simulations. MSc thesis, supervised by
Federico Battiston.
These were the best two years of my life.
In summers, I worked at the Austrian Institute of Technology, and the Complexity Science Hub. I also did a bunch of small projects: helping others set up complicated environments, a Wikidata SPARQL API wrapper in Python for human profile data collection, automated job posting collection and keywords extraction, and so on.
I hold a BSc in Electrical Engineering. In 2024, I combined my engineering and DS knowledge at the Austrian Institute of Technology, working on LiDAR data (as a freelancer).
My Bachelor's thesis: Graph theoretical models and algorithms for portfolio compression.
I also was an intern at Bosch, working in the radar signal processing team.
More: Projects under the projects subpage.
Academia | Industry
Timeline of work and research (WIP)
docx-llm
I am now building docx-llm: a digital product for highly structured Word document generation, using AI agents.
Kulturplanner
Analytics, data engineering, BI for major operahouses, theatres and museums in Europe (contract work).
MSc Thesis
A constructive, causal framework for synergy, redundancy, emergence in dynamical complex networks. Alternative to information theoretic approaches.
AIT LiDAR Freelance
3D LiDAR change detection algorithm evaluation pipeline development end-to-end, and project decision making.
Public Road Network (CEU)
Graph-based OD matrix prediction for Hungarian road networks.
PainterPalette + NetSci
Largest public dataset of painters; NetSci 2025 presentation.
Bosch Radar Intern
Automated radar measurement pipelines and CI/CD tooling.
BSc Thesis
Graph theoretical models for portfolio compression.
Background: Originally, I majored in Electrical Engineering (which may be handy in the future), increasingly becoming more interdisciplinary. Even during those years I focused on discrete mathematics, writing my thesis about applying graph theory on a problem in finance.
Maths guru, especially combinatorics (discrete maths, graphs, algorithms). Competed for ~10 years on the highest levels. Did not qualify for IMO though. Ranked 2nd place multiple times in the TU Budapest mathematics & discrete mathematics competitions.
Free-time: I visit museums, new places I haven't seen, exploring everything from small to big.
Regularly visiting interdisciplinary talks, meetups, workshops.
Also, I love to overthink and overarchitect stuff. I overthink cooking, overorganize my wardrobe and tasks, overthink art, music, society. It's a gift and a burden, but I manage to control it.
Posts
Thoughts and notes
Added soon!: ...
Added soon!