// Data Scientist & ML Engineer · Budapest
I build recommenders, game-playing bots and the small apps around them, and I measure each one against something before I say it works.
§ 01
Rate what you have watched and it finds the separate tastes in your list, then recommends for each one instead of averaging you into a single blur. A Go engine with a React front end, measured on 2,000 held-out users.
A bot that plays Slay the Spire as the Ironclad, talking to the game through CommunicationMod. A learning pipeline trains it; a C++ simulator is the referee, because a bot that misunderstands the rules learns the wrong game.
Daily puzzles with one rule: the server decides. It owns the puzzle, the date and your streak, and a single finished game is enough to keep the streak alive. Next.js on Vercel, Postgres behind it.
Wordle for anime characters: guess the hidden one, and every guess shows how close its series, species, hair, class and age are. About 120 characters, vanilla JavaScript, no framework.
§ 02
A diversity weight shipped at 0.3 for months while ranking worse than no weighting at all, because nothing measured the configuration users were getting. Now the weights are validated at startup and re-measured before any change.
In the audiobook generator a line the model cannot place falls back to the narrator's voice. A hesitant narrator is a small flaw; a confident wrong character is a broken book.
The Slay the Spire bot is checked against a simulator. When the bot's picture of the rules and the simulator's disagree, the bot is the one that is wrong.
§ 03
Third semester abroad, AI and machine learning coursework.
AI specialization at Eötvös Loránd University.
Risk Management Division, Fundamenta Housing Savings Bank - recognized for outstanding performance in risk modeling
Risk Analyst developing internal knowledge systems and database optimization for enhanced data accessibility
§ 04
Python
Go
SQL
Pandas
R
React
JavaScript
Kotlin
Android
Docker
Git
Chrome APIs§ 05
Anidle gives you attribute clues instead of letters. Green is a match, amber is close, and the arrows tell you which way the age lies. This round is a sample; the real one is a click away.
Open the arcade