Research Scientist @ GoogleDeepmind
London
Joined July 2010
- Tweets268
- Following132
- Followers564
- Likes784
Andrey Voynov retweeted
We are hiring research scientists and engineers🧑🔬 on the Gemini Omni team! Both in the US (Mountain View / SF) and Europe (London / Zürich).
Apply here: google.com/about/careers/app…
Andrey Voynov retweeted
We are hiring research scientists in London to build next-generation world models. This is a unique opportunity to be part of a dynamic team.
google.com/about/careers/app…
Andrey Voynov retweeted
🚨 My team at GDM are hiring :) we are looking for people interested in scaling world models, adding new capabilities and using them to solve embodied AGI.
We are looking for people who go above and beyond, put the team first, and believe in our long term mission. Sounds like you? Please apply!!
We have roles in both London and NYC, and hiring for RS and RE. In practice, there's very little difference here as we all get very hands on ;)
Check out how visual diffusion models can effectively solve hard combinatorial geometry problems. Will be presented at #CVPR2026 (highlight ✨ – congrats to all the amazing coauthors @nirgoren @YehezkelShai @OmerDahary @OPatashnik @DanielCohenOr1 )
CVPR 2026 highlight! 🔥
In this work co-led with @YehezkelShai, we show that a plain diffusion model can solve hard geometry problems by treating them as conditional image generation problems. No special architecture needed.
w/ @OmerDahary, @kusichan, @OPatashnik, @DanielCohenOr1
Andrey Voynov retweeted
CVPR 2026 highlight! 🔥
In this work co-led with @YehezkelShai, we show that a plain diffusion model can solve hard geometry problems by treating them as conditional image generation problems. No special architecture needed.
w/ @OmerDahary, @kusichan, @OPatashnik, @DanielCohenOr1
Andrey Voynov retweeted
Video models as Physics simulators. 🌍🎥
[1/] In our latest work, WinDiNet, we finetuned a pre-trained video model into a differentiable physics engine. 1000x faster than traditional CFD solvers.
Project page: rbischof.github.io/windinet_…
Abs: arxiv.org/abs/2603.21210
Check out our most recent work on the training-free video editing from @GoogleDeepMind. One of the cool things about the approach is that it allows modifying the motions of particular scene elements while keeping the dynamics of other parts (check the 🎱 example!). The method is model-agnostic dynaedit.github.io/
Video editing just got more dynamic! 🚀
Thrilled to share DynaEdit: a training-free, text-based method for non-rigid video editing.
Work done during my internship at @GoogleDeepMind with @Roni_Paiss, @kusichan, @inbar_mosseri, @talidekel, @t_michaeli
dynaedit.github.io
Check out our recent paper on motion editing: MotionsV2V. Let’s say you have a video you like, but want some of the objects to behave a bit differently while preserving the rest: for instance, you don't want the cat to look at the fish.
That's exactly what our model does! In this work led by Ryan Burgert we allow v2v edits with trajectories modification. We create a pipeline to generate pairs of videos with partially aligned motion and train a model on it.
Check the details and much more examples on the project page and in the paper!
Thanks and congratulations to the amazing @RyanBurgert @CharlesHerrman8 @forrestercole @ryoo_michael @natanielruizg
web: ryanndagreat.github.io/Motio…
arXiv: arxiv.org/abs/2511.20640
Andrey Voynov retweeted
today we are releasing new research at Google. we tackle the previously unsolved task of editing motion in an existing video. it's called MotionV2V. with it you can move objects in videos, move the camera, and other unprecedented edits in user-provided video
Andrey Voynov retweeted
Add a skeleton, spider, black cat or whatever haunts your imagination straight into your video with object insertion in Flow. Just click on the pencil icon on your video, type out your vision, and watch the story unfold.
Check out our recent paper! TLDR: we learn a very small MLP that dynamically adjust CFG over denoising steps for the best generation.
