Projects
Projects I led based on my own ideas, or where the technical privacy part was my responsibility. These include market products, research tools and public experiments.
2023
-
VeriDome — GDPR-compliant video analysis
2023–present
Pseudonymised face recognition procedure and the video analysis system built on it: creates real value in retail, logistics and industry, and operates in real environments with paying clients. The development was funded by two grants, first the Accelerator programme, then DIMOP Plusz: a total budget of 307.4 million forints (218.2 million forints in support). At Vitarex, this is my project: besides deep technical expertise, I lead research and development, product development, marketing and sales.
2021
-
IML4E — the industrialisation of machine learning
2021–2024
International research and development project focused on MLOps, in which Vitarex, as national lead, developed a scoliosis detection tool for health visitors. The project received the ITEA Award of Excellence in 2025.
-
Anonymisation risk assessment for the Hungarian National Bank
Two studies, totalling about 115 pages: on the privacy issues of face recognition, and on the anonymity of banking transaction data. We also held practical, coding workshops for the bank's experts. The risk assessment methodology developed for transaction data answers whether an anonymisation is strong enough, and if it is broken, how many data subjects could be affected.
2017
-
De-anonymisation simulator
Browser-based demo where our de-anonymisation attack for social networks and the previous method can be tried side by side, on sample datasets, with adjustable parameters. A detailed description helps to understand the details.
-
Browser Extension and Login-Leak Experiment
An experiment at Inria on how browser extensions contribute to web tracking. Preliminary results were presented by invitation at the European Parliament, and after publication, Google also considered making the corresponding change to Chrome.
2016
-
SALab — Structural Anonymity Lab
Java framework (10k+ lines) for studying anonymisation methods for social networks. Contains research-supporting functions and algorithm implementations not available in other frameworks.
-
kmap — making anonymity visible
This is a data visualisation library that helps to understand where and how unique records are in a dataset, i.e. where anonymity may be compromised.
-
webbug.eu
Privacy awareness site about web tracking.
2014
-
SpyTheBug, PrivacyTagger, DocTrack
Attention-grabbing experiments, as originator and project lead for all three. The SpyTheBug Firefox add-on collected statistics on web bugs and showed users how much they were being tracked. PrivacyTagger and DocTrack ran in my university courses: in one, students could choose to sacrifice the anonymity of their own face image for a study bonus; in the other, we tracked the opening of course material with a built-in bug, so we could later discuss hidden surveillance.
2013
-
Mouse heatmap project
A data collection bookmarklet for analysing mouse movement and creating heatmaps. The research that started from this — the possibility of mouse-based identification on the web — later became the topic of several theses.
2012
-
FireGloves
Anti device fingerprinting Firefox add-on that blocks or significantly disrupts browser fingerprinting techniques. Developed by Károly Boda under the supervision of Ádám Földes and myself.
2011
2010
-
EPVP and TracEmail — Email Privacy Visualization Project
2010–2016
Thunderbird and Firefox extensions that show the estimated route of an email on an interactive map, together with the privacy characteristics of the countries involved — that is, who besides the recipient may be able to read it. I led the working group; the project was carried out in cooperation with Privacy International. The updated second version was released as TracEmail, on a new website.
-
System fingerprinting and Cross-Browser Fingerprinting Test 2.0
2010–2012
The first version of the browser-independent fingerprint drew attention to the fact that users can be tracked with an identifier (device fingerprint) that does not need to be stored on their device, as it can be regenerated at any time. The subsequent 2.0 test then showed that users can be tracked without cookies, and even identified independently of the browser: it was run more than 500,000 times and covered by about 50 domestic and international media outlets. My colleagues and I were the first in the world to propose the concept of device identification through the browser, but independently of the browser, in 2011.
2009
2007
-
International PET Portal and Blog
2007–2016
The Hungarian portal for privacy-enhancing technologies, where I was a founding member, developer, moderator, and editor. I wrote more than 350 Hungarian-language posts here; in the years before the GDPR, it was read by 2,500–5,000 unique visitors per month. Most of my attention-grabbing projects, press events, and public programmes were built around the portal — privacy day roundtables, steganography games, Hacktivity sections, radio and TV interviews.