|
Sjoerd van Steenkiste
sjoerdvansteenkiste at gmail dot com
I am a Staff Research Scientist at Google DeepMind in Mountain View.
I currently work on Gemini post-training where I focus on Gemini's agentic reasoning capabilities in real-world domains.
My research interests include:
- Visual scene understanding: learning representations that capture meaningful structure (objects, geometry, etc.) [cf. 1,2]; Controllable generation for scene editing [cf. 3,4].
- LLM reasoning: How data mixtures and architecture affect (pre-)training [cf. 5,6]; Understanding and improving (probabilistic) reasoning [cf. 7,8].
More broadly, I am interested in vision-language models, compositionality, learning
'symbol-like' representations with NNs, and the binding problem.
Previously, I completed my Ph.D. in Computer Science at IDSIA with Jürgen Schmidhuber
and briefly worked as a Postdoctoral Researcher. I received my M.Sc. (2x) and B.Sc. from Maastricht University.
I have also spent time at Google Brain,
NNAISENSE, and AtonRâ.
Academic CV /
Resume /
Google Scholar /
Twitter /
Thesis
|
|