Vivek Veeriah retweeted
My team @GoogleDeepMind is hiring! We are seeking a Research Scientist to rigorously study the internals and generalisation of frontier AI systems, w/ a particular view to AI security.
This is an in-person role at our Platform 37 office in London 🎡
Link to apply in thread 👇
This is an in-person role at our Platform 37 office in London 🎡
Link to apply in thread 👇
7
46
430
39,750
Vivek Veeriah retweeted
I am happy to announce an opening for a student researcher on my team at GDM for 2027, based in London - for a project at the frontier of multi-agent coordination, in duration of 3 months with a possibility of an extension.
5
22
212
15,141
Vivek Veeriah retweeted
It has been a blast hunting cryptids with such a great team 👾
Alongside the Busy Beaver holdout proofs, we also open-sourced smaller 2-symbol Turing machines for three foundational math problems, verified end-to-end in Lean 4 from a blank tape to Mathlib (0 sorry):
- Goldbach's Conjecture: 25 -> 21 states
- Riemann Hypothesis: 120 -> 104 states
- ZF Consistency: 393 -> 380 states
Using Gemini Argon, we compressed the community baseline machines and proved both the state reductions and full mathematical semantics in the kernel: each machine halts if and only if the conjecture fails (or ZF is inconsistent).
Best part of releasing this openly: within days of going live, a community contributor already built on our 104-state proof to push the Riemann machine down to 95 states. 🚀
Huge thanks to everyone across the @GoogleDeepMind team who contributed to this project: @tomzahavy, @robertarail, @dsantosmarco, @vivek_veeriah, @Dahoas1, @zuza777, @MoritzFirsching, Andrew Bolt, Satinder Singh, Hussain Masoom, and Petar Sirkovic, as well as the Busy Beaver community.
Code: github.com/google-deepmin…
Alongside the Busy Beaver holdout proofs, we also open-sourced smaller 2-symbol Turing machines for three foundational math problems, verified end-to-end in Lean 4 from a blank tape to Mathlib (0 sorry):
- Goldbach's Conjecture: 25 -> 21 states
- Riemann Hypothesis: 120 -> 104 states
- ZF Consistency: 393 -> 380 states
Using Gemini Argon, we compressed the community baseline machines and proved both the state reductions and full mathematical semantics in the kernel: each machine halts if and only if the conjecture fails (or ZF is inconsistent).
Best part of releasing this openly: within days of going live, a community contributor already built on our 104-state proof to push the Riemann machine down to 95 states. 🚀
Huge thanks to everyone across the @GoogleDeepMind team who contributed to this project: @tomzahavy, @robertarail, @dsantosmarco, @vivek_veeriah, @Dahoas1, @zuza777, @MoritzFirsching, Andrew Bolt, Satinder Singh, Hussain Masoom, and Petar Sirkovic, as well as the Busy Beaver community.
Code: github.com/google-deepmin…
Over the past year, the Discovery team at @GoogleDeepMind, in a project led by student researcher @dsantosmarco, has made new discoveries in one of the most beautiful open problems at the edge of computation: Finding the 6th Busy Beaver (BB) number.
While most AI math systems roblems, like theorems, this work tackles a different aspect of mathematical creativity: Discovering hidden problems.
For years, the BB Community has pursued one of theoretical computer science’s most captivating collaborative quests: finding the simplest programs that run the longest before halting. Because determining whether these programs ever stop is uncomputable, no universal algorithm can solve it, meaning new mathematical techniques need to be found and invented to make progress.
In this way, the BB challenge serves as an open ended domain for mathematical creativity and, in particular, a generator of hard math problems called Cryptids. These are tiny programs whose halting behavior is governed by and equivalent to deep, unsolved math problems in number theory, similar to the Collatz or Goldbach conjectures. Historically, the BB community had recognized ~20 Cryptids in BB(6), and the best problems were selected for the Busy Math Olympiad (BMO).
Building on their work, Marco built a pipeline powered by Gemini 4 Argon that:
1️⃣ Simulates raw Turing machine traces to compress recurring patterns.
2️⃣ Conjectures mathematical transition rules & runs synthetic experiments to test edge cases.
3️⃣ Proves the equivalence formally in Lean 4. Using this pipeline, Marco and the team discovered 20 new potential Cryptids and verified 33 more machines previously studied in the BB community. All the proofs were open sourced and shared in the community discord channel and we hope that they will help the community to make progress towards resolving BB(6). We also contributed related halting/non-halting proofs directly to the community GitHub repository.
Finding these Cryptids shows hidden links between simple programs and number theory, and helps mathematicians map the limits of computation. I'm also excited about Gemini 4 Argon's ability to help researchers both solve hard problems and accelerate scientific discoveries, of which Busy Beaver is just one example.
Work led by our student researcher @dsantosmarco with paper contributions from @vivek_veeriah, @Dahoas1 , Andrew Bolt, @zuza777 , Satinder Singh, and myself with additional contributions from Hussain Masoom, @maxencefaldor , @MoritzFirsching and Petar Sirkovic.
Read more about it in our paper: arxiv.org/abs/2610.22603
2
4
26
3,005
Incredibly proud to be part of this work on pushing the busy beaver frontier with Argon!
Over the past year, the Discovery team at @GoogleDeepMind, in a project led by student researcher @dsantosmarco, has made new discoveries in one of the most beautiful open problems at the edge of computation: Finding the 6th Busy Beaver (BB) number.
While most AI math systems roblems, like theorems, this work tackles a different aspect of mathematical creativity: Discovering hidden problems.
For years, the BB Community has pursued one of theoretical computer science’s most captivating collaborative quests: finding the simplest programs that run the longest before halting. Because determining whether these programs ever stop is uncomputable, no universal algorithm can solve it, meaning new mathematical techniques need to be found and invented to make progress.
In this way, the BB challenge serves as an open ended domain for mathematical creativity and, in particular, a generator of hard math problems called Cryptids. These are tiny programs whose halting behavior is governed by and equivalent to deep, unsolved math problems in number theory, similar to the Collatz or Goldbach conjectures. Historically, the BB community had recognized ~20 Cryptids in BB(6), and the best problems were selected for the Busy Math Olympiad (BMO).
Building on their work, Marco built a pipeline powered by Gemini 4 Argon that:
1️⃣ Simulates raw Turing machine traces to compress recurring patterns.
2️⃣ Conjectures mathematical transition rules & runs synthetic experiments to test edge cases.
3️⃣ Proves the equivalence formally in Lean 4. Using this pipeline, Marco and the team discovered 20 new potential Cryptids and verified 33 more machines previously studied in the BB community. All the proofs were open sourced and shared in the community discord channel and we hope that they will help the community to make progress towards resolving BB(6). We also contributed related halting/non-halting proofs directly to the community GitHub repository.
Finding these Cryptids shows hidden links between simple programs and number theory, and helps mathematicians map the limits of computation. I'm also excited about Gemini 4 Argon's ability to help researchers both solve hard problems and accelerate scientific discoveries, of which Busy Beaver is just one example.
Work led by our student researcher @dsantosmarco with paper contributions from @vivek_veeriah, @Dahoas1 , Andrew Bolt, @zuza777 , Satinder Singh, and myself with additional contributions from Hussain Masoom, @maxencefaldor , @MoritzFirsching and Petar Sirkovic.
Read more about it in our paper: arxiv.org/abs/2610.22603
0
0
6
210
Vivek Veeriah retweeted
Over the past year, the Discovery team at @GoogleDeepMind, in a project led by student researcher @dsantosmarco, has made new discoveries in one of the most beautiful open problems at the edge of computation: Finding the 6th Busy Beaver (BB) number.
While most AI math systems roblems, like theorems, this work tackles a different aspect of mathematical creativity: Discovering hidden problems.
For years, the BB Community has pursued one of theoretical computer science’s most captivating collaborative quests: finding the simplest programs that run the longest before halting. Because determining whether these programs ever stop is uncomputable, no universal algorithm can solve it, meaning new mathematical techniques need to be found and invented to make progress.
In this way, the BB challenge serves as an open ended domain for mathematical creativity and, in particular, a generator of hard math problems called Cryptids. These are tiny programs whose halting behavior is governed by and equivalent to deep, unsolved math problems in number theory, similar to the Collatz or Goldbach conjectures. Historically, the BB community had recognized ~20 Cryptids in BB(6), and the best problems were selected for the Busy Math Olympiad (BMO).
Building on their work, Marco built a pipeline powered by Gemini 4 Argon that:
1️⃣ Simulates raw Turing machine traces to compress recurring patterns.
2️⃣ Conjectures mathematical transition rules & runs synthetic experiments to test edge cases.
3️⃣ Proves the equivalence formally in Lean 4. Using this pipeline, Marco and the team discovered 20 new potential Cryptids and verified 33 more machines previously studied in the BB community. All the proofs were open sourced and shared in the community discord channel and we hope that they will help the community to make progress towards resolving BB(6). We also contributed related halting/non-halting proofs directly to the community GitHub repository.
Finding these Cryptids shows hidden links between simple programs and number theory, and helps mathematicians map the limits of computation. I'm also excited about Gemini 4 Argon's ability to help researchers both solve hard problems and accelerate scientific discoveries, of which Busy Beaver is just one example.
Work led by our student researcher @dsantosmarco with paper contributions from @vivek_veeriah, @Dahoas1 , Andrew Bolt, @zuza777 , Satinder Singh, and myself with additional contributions from Hussain Masoom, @maxencefaldor , @MoritzFirsching and Petar Sirkovic.
Read more about it in our paper: arxiv.org/abs/2610.22603
While most AI math systems roblems, like theorems, this work tackles a different aspect of mathematical creativity: Discovering hidden problems.
For years, the BB Community has pursued one of theoretical computer science’s most captivating collaborative quests: finding the simplest programs that run the longest before halting. Because determining whether these programs ever stop is uncomputable, no universal algorithm can solve it, meaning new mathematical techniques need to be found and invented to make progress.
In this way, the BB challenge serves as an open ended domain for mathematical creativity and, in particular, a generator of hard math problems called Cryptids. These are tiny programs whose halting behavior is governed by and equivalent to deep, unsolved math problems in number theory, similar to the Collatz or Goldbach conjectures. Historically, the BB community had recognized ~20 Cryptids in BB(6), and the best problems were selected for the Busy Math Olympiad (BMO).
Building on their work, Marco built a pipeline powered by Gemini 4 Argon that:
1️⃣ Simulates raw Turing machine traces to compress recurring patterns.
2️⃣ Conjectures mathematical transition rules & runs synthetic experiments to test edge cases.
3️⃣ Proves the equivalence formally in Lean 4. Using this pipeline, Marco and the team discovered 20 new potential Cryptids and verified 33 more machines previously studied in the BB community. All the proofs were open sourced and shared in the community discord channel and we hope that they will help the community to make progress towards resolving BB(6). We also contributed related halting/non-halting proofs directly to the community GitHub repository.
Finding these Cryptids shows hidden links between simple programs and number theory, and helps mathematicians map the limits of computation. I'm also excited about Gemini 4 Argon's ability to help researchers both solve hard problems and accelerate scientific discoveries, of which Busy Beaver is just one example.
Work led by our student researcher @dsantosmarco with paper contributions from @vivek_veeriah, @Dahoas1 , Andrew Bolt, @zuza777 , Satinder Singh, and myself with additional contributions from Hussain Masoom, @maxencefaldor , @MoritzFirsching and Petar Sirkovic.
Read more about it in our paper: arxiv.org/abs/2610.22603
30
76
531
50,493
Vivek Veeriah retweeted
ttrl was the silver bullet in alphaproof and before that in chess
ssi are ploughing the fields of a new paradigm shift, TTT (test time training). it will allow the models to update a subset of their weights on the fly.
so your digital co worker won’t forget everything about your business needs in each new chat, continual learning is here. ls a scratch pad to think and allocating test time compute, ilya has found the next shift and pushed hard.
here’s a cute silly image to show you the third axis we are scaling.
the laws. they hold 🥹
0
1
9
1,722
Vivek Veeriah retweeted
We'll be demoing our work on COrigami at the @Google booth next week!
📄✨ 📅 Wed, July 8 ⏰ 12:30 PM — 1:30 PM 📍 Kiosk #2
We've also made some crease patterns available for you to try out: tomzahavy.com/_files/archive…
📄✨ 📅 Wed, July 8 ⏰ 12:30 PM — 1:30 PM 📍 Kiosk #2
We've also made some crease patterns available for you to try out: tomzahavy.com/_files/archive…
I am excited to share our latest work from the Discovery team at @GoogleDeepMind: COrigami, an end-to-end pipeline for co-designing origami with Gemini!
Origami is a unique mix of math, art and design. Creating origami involves turning abstract concepts into real-world objects, foldability. To tackle this, COrigami calls Gemini to generate a semantic stick figure—an abstract JSON code—refined through a visual feedback loop. It then calls custom packing, solving, shaping, and simulation tools. Driven by another self-improving reinforcement learning (RL) loop, the system produces visually recognisable models represented as SVG crease patterns.
Despite scarce data availability, this approach demonstrates how combining RL with frontier models like Gemini can assist human creativity and produce physical art. The generated patterns serve as mathematically grounded starting points for origami artists to fold and shape into a final, physical design. The models below were produced by our system and then folded and shaped by @brandon_w0ng.
Read the full paper here: arxiv.org/abs/2606.26299
1
4
18
10,612
Vivek Veeriah retweeted
I am excited to share our latest work from the Discovery team at @GoogleDeepMind: COrigami, an end-to-end pipeline for co-designing origami with Gemini!
Origami is a unique mix of math, art and design. Creating origami involves turning abstract concepts into real-world objects, foldability. To tackle this, COrigami calls Gemini to generate a semantic stick figure—an abstract JSON code—refined through a visual feedback loop. It then calls custom packing, solving, shaping, and simulation tools. Driven by another self-improving reinforcement learning (RL) loop, the system produces visually recognisable models represented as SVG crease patterns.
Despite scarce data availability, this approach demonstrates how combining RL with frontier models like Gemini can assist human creativity and produce physical art. The generated patterns serve as mathematically grounded starting points for origami artists to fold and shape into a final, physical design. The models below were produced by our system and then folded and shaped by @brandon_w0ng.
Read the full paper here: arxiv.org/abs/2606.26299
Origami is a unique mix of math, art and design. Creating origami involves turning abstract concepts into real-world objects, foldability. To tackle this, COrigami calls Gemini to generate a semantic stick figure—an abstract JSON code—refined through a visual feedback loop. It then calls custom packing, solving, shaping, and simulation tools. Driven by another self-improving reinforcement learning (RL) loop, the system produces visually recognisable models represented as SVG crease patterns.
Despite scarce data availability, this approach demonstrates how combining RL with frontier models like Gemini can assist human creativity and produce physical art. The generated patterns serve as mathematically grounded starting points for origami artists to fold and shape into a final, physical design. The models below were produced by our system and then folded and shaped by @brandon_w0ng.
Read the full paper here: arxiv.org/abs/2606.26299
17
100
434
72,669
Vivek Veeriah retweeted
Our team just released DiffusionGemma: a new state-of-the-art open diffusion LLM 🚀
Pushing the RL research for DiffusionGemma has been fascinating. RL doesn't just boost performance, it unlocks blistering speedups (>1000 tok/s on a single H100) ⚡️
Pushing the RL research for DiffusionGemma has been fascinating. RL doesn't just boost performance, it unlocks blistering speedups (>1000 tok/s on a single H100) ⚡️
DiffusionGemma is our new experimental open model with up to 4x faster output on dedicated GPUs.
Instead of predicting word-by-word, it generates entire blocks of text simultaneously. This lets the model self-correct and format complex markdown in real time.
5
10
52
3,714
Vivek Veeriah retweeted
Talk to us about this work at #NeurIPS2025:
Main track: Fri, 11-14, #311
Creative AI track: Wed, 11-14, 29A-D
GDM booth talk, Wed, 16:30-17
GDM booth demo, Fri, 9-12
sites.google.com/view/puzzle-gen
@GoogleDeepMind #NeurIPS2025 #creativity #Chess
Main track: Fri, 11-14, #311
Creative AI track: Wed, 11-14, 29A-D
GDM booth talk, Wed, 16:30-17
GDM booth demo, Fri, 9-12
sites.google.com/view/puzzle-gen
@GoogleDeepMind #NeurIPS2025 #creativity #Chess
I am excited to share a work we did in the Discovery team at @GoogleDeepMind using RL and generative models to discover creative chess puzzles 🔊♟️♟️ #neurips2025
🎨While strong chess players intuitively recognize the beauty of a position, articulating the precise elements that emains elusive. To address this, we pre-trained generative models on public datasets and then applied reinforcement learning, using novel rewards designed for uniqueness, counter-intuitiveness, realism, and novelty. This approach doubled the number of novel chess puzzles compared to the original training data, while successfully maintaining aesthetic diversity.
Three distinguished experts—International Master of chess compositions Amatzia Avni (author of "Creative Chess"), Grandmaster Jonathan Levitt @JonathanLevitt7 (author of "Secrets of Spectacular Chess"), and Grandmaster Matthew Sadler @gmmds (author of "Game Changer")—evaluated and selected the puzzles they found most compelling. Their preference was for puzzles exhibiting original, paradoxical, surprising, and naturally occurring positions, with particular emphasis on those that integrated aesthetic themes in innovative ways and demonstrated exceptional over-the-board vision.
🧩Try to solve the puzzles @chesscom: chess.com/c/2wCTN7Uv2
0
6
32
12,100
Vivek Veeriah retweeted
Unfortunately I cannot make this NeurIPS because of the visa issue. But we will still have @TZahavy, @vivek_veeriah, and @tomtumiel presenting our chess puzzle generation paper 👇. Feel free to talk to them or try the puzzles by yourself!
sites.google.com/view/puzzle-gen
#NeurIPS2025
sites.google.com/view/puzzle-gen
#NeurIPS2025
Talk to us about this work at #NeurIPS2025:
Main track: Fri, 11-14, #311
Creative AI track: Wed, 11-14, 29A-D
GDM booth talk, Wed, 16:30-17
GDM booth demo, Fri, 9-12
sites.google.com/view/puzzle-gen
@GoogleDeepMind #NeurIPS2025 #creativity #Chess
1
5
28
5,516
In San Diego for #NeurIPS2025 until Sunday to present our work on Generating Creative Chess Puzzles.
If you're interested in the intersection of AI and creativity, let's chat.
We are also looking for SRs to push the frontiers of Creativity, DM me!
If you're interested in the intersection of AI and creativity, let's chat.
We are also looking for SRs to push the frontiers of Creativity, DM me!
Talk to us about this work at #NeurIPS2025:
Main track: Fri, 11-14, #311
Creative AI track: Wed, 11-14, 29A-D
GDM booth talk, Wed, 16:30-17
GDM booth demo, Fri, 9-12
sites.google.com/view/puzzle-gen
@GoogleDeepMind #NeurIPS2025 #creativity #Chess
1
1
12
1,670
Vivek Veeriah retweeted
📣 Hiring Alert: @GoogleDeepMind Student Researcher - 2026
@Xidong_Feng and I are hiring a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧 to investigate Artificial Curiosity and Intrinsic Motivation algorithms in general discovery 🔭
Apply below! 👇🏻:
@Xidong_Feng and I are hiring a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧 to investigate Artificial Curiosity and Intrinsic Motivation algorithms in general discovery 🔭
Apply below! 👇🏻:
4
21
229
25,263
Vivek Veeriah retweeted
📣 Hiring Alert:@GoogleDeepMind Student Researcher - 2026 Spring
@FaccioAI and I are hiring a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧to investigate Artificial Curiosity and Intrinsic Motivation algorithms in general discovery 🔭.
For a sense of the check out our previous project: arxiv.org/abs/2510.23881.
Apply below! 👇🏻:
@FaccioAI and I are hiring a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧to investigate Artificial Curiosity and Intrinsic Motivation algorithms in general discovery 🔭.
For a sense of the check out our previous project: arxiv.org/abs/2510.23881.
Apply below! 👇🏻:
8
27
218
40,614
Vivek Veeriah retweeted
This announcement feels personal. I had just joined the maths team and the first thing I was told: "We're 3 months away from the IMO."
While I was only there for the finish line, this was the culmination of years of the team's foundational work. So privileged to have been there!
While I was only there for the finish line, this was the culmination of years of the team's foundational work. So privileged to have been there!
Last year, AlphaProof & AlphaGeometry reached a key landmark in AI by achieving silver medal level performance at the International Math Olympiad.
Today, @Nature is publishing the methodology behind our amazing agent AlphaProof! @GoogleDeepMind
Paper: nature.com/articles/s4158…
0
1
20
3,774
Are you excited about pushing the frontiers of Creativity in language models, especially for scientific discovery?
We are hiring a Student Researcher at @GoogleDeepMind to work with us on this mission 🚀
We are hiring a Student Researcher at @GoogleDeepMind to work with us on this mission 🚀
📣 Hiring Alert: Student Researcher - 2026
@vivek_veeriah and I are looking for a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧!
We will be investigating how creativity in LLMs generalizes, with application to scientific discovery 🔭
Apply below! ⬇️
1
7
87
9,126
Vivek Veeriah retweeted
📣 Hiring Alert: Student Researcher - 2026
@vivek_veeriah and I are looking for a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧!
We will be investigating how creativity in LLMs generalizes, with application to scientific discovery 🔭
Apply below! ⬇️
@vivek_veeriah and I are looking for a PhD Student Researcher to join the GDM Discovery team in London 🇬🇧!
We will be investigating how creativity in LLMs generalizes, with application to scientific discovery 🔭
Apply below! ⬇️
9
33
295
40,920
Vivek Veeriah retweeted
New from me, in @Nature: Mathematicians put AI model AlphaProof to the test. A solicited News & Views article about @GoogleDeepMind AlphaProof that was an absolute joy to write!
rdcu.be/ePz4Q
rdcu.be/ePz4Q
7
27
152
29,849
Vivek Veeriah retweeted
Last year, AlphaProof & AlphaGeometry reached a key landmark in AI by achieving silver medal level performance at the International Math Olympiad.
Today, @Nature is publishing the methodology behind our amazing agent AlphaProof! @GoogleDeepMind
Paper: nature.com/articles/s4158…
Today, @Nature is publishing the methodology behind our amazing agent AlphaProof! @GoogleDeepMind
Paper: nature.com/articles/s4158…
7
74
423
69,990
Vivek Veeriah retweeted
Very excited that our AlphaProof paper is finally out!
It's the final thing I worked on at DeepMind, very satisfying to be able to share the full details now - very fun project and awesome team!
julian.ac/blog/2025/11/1…
It's the final thing I worked on at DeepMind, very satisfying to be able to share the full details now - very fun project and awesome team!
julian.ac/blog/2025/11/1…
20
96
1,005
101,249