We collected preferences in a form of 50k comparisons of youtube videos from ~2k people out of which 500 are affiliated with research institutions. The results were processed in real time with our byzantine-proof algorithms and are available for the general public in our browser extension. The analysis is available as a whitepaper. The project is with a bigger team at EPFL now

Working on better defenses against adversarial policies in multi-agent reinforcement learning at the Center for Human-Compatible AI at the University of California, Berkeley. We wanted to find defenses and instead found much more adversarial policies... :)

Working at Google Research on Causal Models in Reinforcement Learning. We decided to work on uncovering the causal graph of an environment, and to deal with spurious correlations we propose to do interventions. The project resulted in a paper at ICLR CLRM workshop

- 2017-2018: working on an augmented reality & robots for education. the AR library is on GitHub, research paper pending after a run at the International School of Geneva - 2018-2019: working on fault tolerance of neural networks. Paper is on arXiv and the project was presented at the DS3 summer school.

- we provide software solutions for escape rooms - we allow for cloud-based remote control of the escape room - all hardware is remotely controlled - about 40 quests used our framework - project sold to one of our clients

Extended the algorithm to search for non-convexities in images of quadratic maps to the case when the number of them is infinite. The paper is on arXiv and the project was presented at a summer school

Backend engineering with TypeScript and ArangoDB-based database