Veniamin Veselovsky

CEO and Co-Founder at Ludus Labs. On leave from PhD in CS at Princeton. Research @ dlab and css lab

prof_pic.jpg
venia[at]veselovsky[dot]org

Hello 👋!

Currently focused on making AIs compete against each other real physically-simulated sports.

Before this I spent time applying/studying ML models to better understand society and individuals. I was a PhD student (on leave) in CS at Princeton working with Arvind Narayanan, Matt Salganik, and Tom Griffiths supported by the Gordon Wu Fellowship. Prior to coming here I spent a few years working with Robert West and dlab at EPFL. Additionally, I spent 6 months at the Max Planck Institute in Saarbrücken analyzing internet bytes with Anja Feldmann and Ingmar Weber. I got my undergrad in Toronto, Canada studying math and international relations. The last two of the years I had the honour of working with Ashton Anderson, Robert Soden, and Robert Bothwell!

selected publications

2024

  1. Do Llamas Work in English? On the Latent Language of Multilingual Transformers
    Chris Wendler, Veniamin Veselovsky, Giovanni Monea, and 1 more author
    arXiv preprint arXiv:2402.10588, 2024
  2. Evaluating Language Model Agency through Negotiations
    Tim R. Davidson, Veniamin Veselovsky, Martin Josifoski, and 4 more authors
    In ICLR, 2024

2023

  1. Artificial Artificial Artificial Intelligence: Crowd Workers Widely Use Large Language Models for Text Production Tasks
    Veniamin Veselovsky, Manoel Horta Ribeiro, and Robert West
    arXiv preprint arXiv:2306.07899, 2023
  2. Reddit in the Time of COVID
    Veniamin Veselovsky, and Ashton Anderson
    In Proceedings of the International AAAI Conference on Web and Social Media, 2023

2021

  1. Imagine all the people: Characterizing social music sharing on reddit
    Veniamin Veselovsky, Isaac Waller, and Ashton Anderson
    In Proceedings of the International AAAI Conference on Web and Social Media, 2021