Research scientist @DeepMind and lecturer @UCBerkeley. Interested in AI/education/visual art/nature. Previously @OpenAI and @UCBerkeley.
Joined April 2020
- Tweets 275
- Following 237
- Followers 5,653
- Likes 1,093
44 Photos and videos
Igor Mordatch retweeted
May 28
Our team at DeepMind is growing.
We've assembled a world class physics material science team and are building an experimental lab.
If you want to solve real problems at the intersection of AI material science to unlock a technological revolution - this is the place. Apply 👇
48
76
1,444
4,316,658
Igor Mordatch retweeted
18 Dec 2025
AI has the potential to compress the time needed for new discoveries from years to days. 📉
That’s why we’re supporting the US Dept. of @ENERGY's Genesis Mission – providing National Labs with access to AI tools to help accelerate research in physics, chemistry, and beyond. → goo.gle/4s0Bujw
58
110
831
268,065
Igor Mordatch retweeted
23 Nov 2025
Gemini Nano Banana Pro can solve exam questions *in* the exam page image. With doodles, diagrams, all that.
ChatGPT thinks these solutions are all correct except Se_2P_2 should be "diselenium diphosphide" and a spelling mistake (should be "thiocyanic acid" not "thoicyanic")
:O
324
934
8,827
3,006,809
25 Nov 2025
Excited to be at NeurIPS in San Diego from Thursday to Saturday next week! If you want to chat about AI for accelerating science (esp. materials), multi-modality, robotics, or intersections of these things, please reach out!
2
1
93
7,164
18 Nov 2025
Excited that Gemini 3 is finally out! Visual perception is particularly improved in this version and I'm excited to see all the downstream effects of that.
18 Nov 2025
We’ve been intensely cooking Gemini 3 for a while now, and we’re so excited and proud to share the results with you all. Of course it tops the leaderboards, including @arena, HLE, GPQA etc, but beyond the benchmarks it’s been by far my favourite model to use for its style and depth, and what it can do to help with everyday tasks.
3
680
17 Nov 2025
Humans benefit from visualizing data. Do AI models need visualization too, or can they just take in raw data text? A fun study co-made with Claude to answer this question. Conclusion: it seems like AI currently benefits from looking at data visualizations! mordatch.github.io/posts/dat…
2
4
18
2,116
17 Nov 2025
See blog post for more details mordatch.github.io/posts/dat… View and download the dataset of csv, visualizations, and questions to try it yourself here: huggingface.co/datasets/imor…. And here is a Claude transcript for the entire project: claude.ai/share/ca4da6ab-189…
1
1
403
17 Nov 2025
Working on this with Claude in a single long session was really fun and very fast. What could have been a summer research project was done in a few weekend bursts on the side. I'm excited to try looking into more research questions in this manner!
2
341
7 Oct 2025
Personally, I love this plot because it so crisply shows the value of active (RL) vs passive (SFT) experience for embodied agents: just 1% of active (RL) interaction gives you jump from orange to blue which you can't approach by just pouring in more passive SFT data (orange).
7 Oct 2025
Super excited to finally share our work on “Self-Improving Embodied Foundation Models”!!
(Also accepted at NeurIPS 2025)
• Online on-robot Self-Improvement
• Self-predicted rewards and success detection
• Orders of magnitude sample-efficiency gains compared to SFT alone
• Generalization enables novel skill acquisition
🧵👇[1/11]
1
2
19
4,016
7 Oct 2025
Awesome work by @coolboi95 Kamyar and the team and I can't wait to see what you achieve at Generalist!
1
4
658
Igor Mordatch retweeted
19 Sep 2025
Self-Improving Embodied FMs
• 2-stage recipe: SFT online RL w/ self-predicted rewards
• Boosts sample efficiency: 10% robot time → 45%→75% success (vs. 8× data → only 60%)
• Unlocks autonomous skill acquisition beyond imitation data
6
33
241
18,182
29 Aug 2025
Pointing and visual guidance - first shown in Astra - now available in Gemini Live! I love this as a path to for interaction between AI and the physical world complementary to robotics. blog.google/products/gemini/…
1
5
811
27 Aug 2025
Nice to see visual overlays also being adopted in Grok!
27 Aug 2025
BREAKING 🚨: xAI launched objects highlighting on Grok for iOS!
Object highlighting feature can point to certain objects in Grok Vision mode.
Grok got Astra before Gemini 🤯
4
1,345
29 May 2025
Another example of Astra's non-verbal pointing and gestures can be helpful in tutoring situations: youtube.com/watch?v=MQ4JfafE…
3
1,372
20 May 2025
Really excited to share what I and many amazing colleagues have been working on recently. Especially highlighting as a way to non-verbally communicate and help focus user's attention. So much more is possible!
20 May 2025
Last year, we introduced Project Astra: a research prototype exploring capabilities for a universal AI assistant. 🤝
We’ve been making it even better with improved voice output, memory and computer control - so it can be more personalized and proactive. Take a look ↓ #GoogleIO
11
1,442
13 Jan 2025
Attempts at model self-improvement often lead to degeneration over time. But co-improving multiple models lets us maintain diversity and avoid degradation. Hopefully this can be a step towards AI systems with long-running self-improvement capabilities!
Introducing multi-agent self-improvement with LLMs!
llm-multiagent-ft.github.io/
Instead of self improving a single LLM, we self-improve a population of LLMs initialized from a base model.
This enables consistent self-improvement over multiple rounds.
2
3
58
6,224
11 Dec 2024
Really excited about the possibilities with this!
11 Dec 2024
The Multimodal Live API lets you create interactive experiences that respond in real-time to text, audio, and video input.
1
9
1,811