I’m a staff research scientist with DeepMind, working with Graph Neural Networks and RL. I’m particularly interested in the intersection of physics simulation and machine learning.
My background is in physically-based simulation for computer graphics. Before joining DeepMind, I developed garment simulation tech for Avametric, a SF-based started for virtual fit (now aquired by Gerber). In academia, I worked on fracture and thin shell mechanics with James O'Brien at UC Berkeley as a postdoc, and studied turbulent fluid simulation for graphics at ETH Zurich in Markus Gross‘ group as a PhD student. In ancient times I worked in environmental physics at IUP Heidelberg. I co-developed the simulation frameworks Arcsim and Mantaflow, which are used in both research and industry.
Sometimes I make games. I’ve worked with the excellent folks at DoubleFine and Shinyshoe, and have a few of projects of my own.
Stingrays are awesome.
Contact
tob.pfaff (at) gmail.com