Women in Data Science
Budapest, 2025

Join us on our annual Women in Data Science conference on October 9, 2025 at Corvinus Gellert Campus.

Don’t miss out – register for free today!

WiDS Budapest is an independent event organized by Orsolya Vasarhelyi, Andrea Sipos, Veronika Hamar, Melanie Oyarzún, Yajie Wang, Amina Akhmadiyeva and Akbota Saduakassova to to be part of the mission to increase participation of women in data science and to feature outstanding women doing outstanding work

About WiDS Global

Speakers 2025

Katalin Fehér

Associate Professor
Ludovika University of Public Service

Anikó Hannák

Professor of Social Computing
University of Zurich

Adrienn Juhász

Lead Data Scientist
Mastercard, Ekata

Noémi Ligeti-Nagy

Research Fellow
ELTE Research Centre for Linguistics

Judit Mokos

Technical Instructor
Datapao

Judit Nagy

Vice-Rector, Associate Professor
Corvinus University of Budapest

Rebeka O’Szabó

Assistant Professor
Corvinus University of Budapest

Viola Petra

Engineering Director
Google

Zoé Rimay

Wealth Management Chief Data Office Budapest Lead
Morgan Stanley

Krisztina Vámos

Chief Data Officer
City of Budapest

Conference Schedule

October 9, 2025

16:00–16:05
Welcome to WiDS Budapest 2025!
Veronika Hamar
WiDS Ambassador; Executive Director, Center for Collective Learning

16:05–16:15
Opening Remarks
Judit Nagy
Vice-Rector of Student Affairs, Corvinus University of Budapest

16:15–16:35
HuGME: How we measure Hungarian in generative models
Noémi Ligeti-Nagy
Research Fellow, ELTE Research Centre for Linguistics

16:40–17:00
The behavioral blueprint of high-performing teams: lessons from escape rooms
Rebeka O’Szabó
Assistant Professor, Corvinus University of Budapest

17:00–17:20
Mind the gap – between academic and industrial data science
Judit Mokos
Technical Instructor, Datapao

Abstract

I began my career in academia as a biologist researcher who somehow always became “the data person” in every team. At first, it was just an informal label, but over time, people started calling me a data scientist. So when I decided to step outside the academic bubble, it felt natural to apply for data scientist roles. I eventually landed at a data consultancy firm as a technical instructor, which is a fancy way of saying I teach people about data.

What I didn’t realise was that moving from academic data science to corporate data science wouldn’t be a neat little step, it would be a giant leap across a very wide gap. The role I knew and practised in research bore little resemblance to what a data scientist does in industry. It wasn’t just about different workplaces, it felt like stepping into an entirely different profession. The new world came with its own language, tools, software and logic, starting with what “being a data scientist” even means.

I realised that a university researcher or lecturer and an industry data scientist would often struggle to speak the same language, even though they could learn enormously valuable lessons from one another.

This talk will explore these differences in how two worlds use the same title for two very different jobs. I hope to make this divide feel less like a yawning chasm and more like a manageable gap: one you simply need to mind.

17:20–17:40
Building Trust from Messy Data: How We Transform Features to Catch Fraud
Adrienn Juhász
Lead Data Scientist, Mastercard, Ekata

Abstract

The rapid growth of online shopping has come with a significant increase in online fraud. By leveraging behavioural, personal, and device insights from past transactions, we’re training cutting-edge machine learning models to predict and prevent fraudulent activities. With over 30 million real-time transactions processed daily through our system, the focus is not only on scalability but also on developing low-latency solutions.

The presentation will begin by providing an overview of the vast amount of data collected from online transactions and outlining the key data challenges encountered. From there, I will show some powerful feature engineering techniques that transform messy inputs into meaningful signals for modeling. During the presentation I will address important questions such as:

– Why is a balanced dataset essential?
– What dangers arise from improper data cleaning?
– How do we handle different patterns across countries?

After the talk, it will be clear that more data is not always the better choice, and why it’s essential to understand the data thoroughly before building any model with it.

17:40–17:45
The First WiDS Budapest Datathon Highlights
Melanie Oyarzún
WiDS Ambassador, Postdoctoral Research Fellow CCL, Corvinus University of Budapest

17:45–18:15
Panel Discussion: Ethical and Responsible Use of Data Science/AI in Policy, Governance and Business
Moderator: Katalin Fehér
Associate Professor, Ludovika University of Public Service

PANELISTS

Anikó Hannák
Professor of Social Computing,
University of Zurich

Zoé Rimay
Wealth Management Chief Data Office Budapest Lead, Morgan Stanley

Viola Petra
Engineering Director, Google

Krisztina Vámos
Chief Data Officer, City of Budapest

18:15-18:20
Closing
WiDS Budapest team

18:20-20:00
NETWORKING & POSTER SESSION

Poster Session

October 9, 2025

18:20-20:00
Smarter Screening: Statistical and Machine Learning Approaches for Down’s Syndrome Risk Prediction
Pauline Ang’ang’o

Can AI Replace a Child? Data Science and the Future of Intergenerational Support
Xiaoru Lin

Evaluating Large Language Models for Gender Bias in Academic Knowledge Production
Judit Hermán, Kíra Diána Kovács, Yajie Wang and Orsolya Vásárhelyi

Detection over Deception—An Inverted Turing Test with LLMs
Manran Zhu, Zsófia Hajnal and Orsolya Vásárhelyi

“Pick-up doesn’t start with booking—it starts with search.” A Data-Driven Hypothesis Test and its Operational Impact in Hotel Sales Marketing
Kriszta Kozma-Renge

Friendship Moderates Hierarchy-Based Cooperation in Childhood
Melanie Oyarzún, Carlos Rodriguez-Sickert, and Cristian Candia