Research Scientist

A short and urgent vacancy for mathematicians, physicists, neuroscientists, and others.

Remember PIBBSS — a non-profit institute that pays scientists stipends to work on, for example, AI interpretability? We are again looking for full-time Research Scientists for them! 🧪

The task is to use percolation theory from statistical physics to build a model of how training data is structured — to understand how its structure affects learned features.

The submission deadline is June 21st — the end of this week. But if you are very interested and absolutely cannot make it, let us know, and we'll figure something out!

Let's be brief: 📖 PhD or its equivalent in a technical discipline; 🔮 experience with DL and neural networks; 💬 ability to clearly communicate complex ideas to people; 🧡 motivation to make AI a little more understood and safer.

May also be useful: cool publications (in ML and beyond), knowledge of percolation theory and other statistical physics, stochastic processes (especially coalescent theory), combinatorics, fractal geometry, neural scaling laws. And, of course, mechanistic interpretability and any courses or camps on AI safety will also be relevant!

They pay approximately $100k-$250k per year, remote — but the whole team gets together 2-4 times a year, so you'll have time to get to know your colleagues.

Dasha @d_shteinardt 💞 is ready to process resumes rapidly, answer any questions, and carefully guide you through the process all week long.