Sr. Staff RS GenAI @GoogleDeepMind. 🇯🇵-born 🇨🇳🇨🇦. ex: Omni Thinking, Gemini 1.5-3, GPT-4 @OpenAI, GoogBrain Robotics (JP: @shanegJP). Personal opinions.

Mountain View, CA
Joined June 2016
I had the pleasure of catching up with Geoff Hinton yesterday to celebrate his birthday. As we took a long, winding walk along the coast, I was reminded of: not all tokens are born equal. Just as his insights and words from a decade ago still trigger new chains of thought for me, yesterday’s conversation was exceptionally fruitful. We spent some time discussing Tim Lillicrap, my current manager and co-lead of Gemini thinking, whose research on random matrix feedback from ten years ago actually inspired my very first paper with Ilya Sutskever. Geoff offered his perspective on exactly why that method worked, dissecting the phenomenon with precise intuitions grounded in mathematics. Our discussion ranged from the technical to the philosophical. He remains as intellectually open-minded as ever, still imagining for true novel breakthroughs that go beyond the current paradigms of backpropagation. Our last (brief) chat was shortly after his foward-forward algorithm in late 2022. We also explored global perspectives, touching on the dynamics between China, Japan, and the West. We discussed how the philosophical roots of Asian cultures might be better suited for the societal shifts of a post-AGI world, and what the West could learn from the East to ensure that the future is positive. On a personal note, I was touched by his genuine concern for the next generation. When I mentioned my daughter, he agreed that raising children is the ultimate way to reshape one’s priorities--a grounding force that liberates you from immediate 'capitalist FOMO' and refocuses you on the long term. His words carry the weight of years of high-quality processing; they aren't just outputs of raw intelligence, but of deep, sincere care for humanity. It is clear he cares about far more than any single technology or nation. His reasoning and stories serve as a powerful reminder to understand and explore the broader world—proof that a vast perspective is what shapes greater wisdom and enhances our humanity. I attended NeurIPS 2024 hoping to congratulate him on his Nobel Prize but missed the chance. I am so glad I was finally able to do so this year on his birthday. And finally, he left me with a few more important predictive tokens: 'Google and Gemini will win' :)"
12
31
5
548
124,311
I’m starting a new early-stage AI fund, doing what I love: partnering with founders from company formation on bold bets, long before consensus. The fund’s LPs will include Sequoia, Jeff Dean, Sachin Katti, Philip Wong (my Stanford professor and Chief Scientist adviser to TSMC), and others whose work has shaped our industry and how I think about AI’s future. I’ll remain a partner at Sequoia and keep serving on my boards on Sequoia’s behalf. Sequoia has been my home for the past 11 years. I joined as the youngest investor the firm had ever hired. I grew up here. Through years of intense debate and shared work, I found lifelong mentors and friends who took me under their wing and taught me this craft. They taught me to aim high and take big swings. I’ll carry that into what I build: outsized ambition, relentless truth-seeking and curiosity, and a refusal to become complacent. Working alongside founders has made me want to build something of my own: a firm shaped from day one around the convictions I’ve developed over the past decade. Life has taught me not to take time for granted. This is how I want to spend mine. Why now We’re seeing a new way of building companies in today’s AI era. Research and products often advance together. A breakthrough can open a market or accelerate a business far faster than expected. Some companies require extraordinary capital and compute early on. Others reach meaningful scale with remarkably small teams. The rules for how big these businesses can get, and how fast, are being rewritten. But early on, the technology is unproven, the resources required enormous, and the vision seemingly implausible. This is what I love most: seeing the potential right from the beginning, and helping founders realize their full scale of ambition. More than ever, it starts with an extreme willingness to believe early, with bold, independent conviction in a founder’s technical insight before any product or market validation. Then, it means helping founders take the biggest swings, giving the rare, long-term patience to see those bets through uncertainty, ambiguity, and turbulence, and serving as a level head when the stakes are highest. I learned this kind of partnership at Sequoia. It’s the foundation of the firm I’m building. Believing before consensus My fascination with AI’s potential took root at Stanford in the early-to-mid 2010s, studying computer science and working in Andrew Ng’s lab. The friendships I formed connected me to a community whose people and ideas would help shape modern AI and build some of our generation’s defining companies. What stayed with me was the gap between what those people saw was possible and what the wider world could imagine. I’m drawn to founders pursuing ambitions that sound impossible today. I start with a simple question: “What if everything goes right?” Then we work backward together to make it happen. In 2023, Deepak and Abhinav set out to build a general-purpose robot brain at @SkildAI, before physical AI was a consensus bet. Their sim-to-real breakthroughs gave me conviction where earlier robotics approaches had fallen short. But the data flywheel required deployment at scale. We made commercialization as urgent as research and brought in strategic partners like NVIDIA and SoftBank to support the ambition. Three years in, Skild is advancing the frontier of physical AI with nine figures in annualized revenue. In 2024, Ioannis and Misha founded @reflection_ai with an insight into scaling reinforcement learning. As that grew into the ambition to build a Western frontier lab for open models, the hard question was how a startup competes with labs spending tens of billions. An introduction to NVIDIA and many compute calls and recruiting events later, the team has built Beam, now one of the leading US open models. In 2025, Anna and Azalia, the co-creators of AlphaChip, founded @RicursiveAI to design models and hardware together. From day one, we made early customers and ecosystem partners a priority. Today, they are showing order-of-magnitude improvements on real workloads, working with some of the industry’s most consequential players, and reaching a defining inflection point. The ambition keeps growing as these companies scale. @linear has always had a kind of magic to it. Now it’s the system a new generation of AI companies builds on, for teams and agents alike. @tavus saw early that we’d one day talk to AI the way we talk to each other. Today, it’s building the human computing interface for a world of AI colleagues. @Mach_Industries began as hydrogen experiments in an MIT dorm room. It has since become one of the most exciting next-generation defense companies. Making the ambition possible Being a first believer is only the beginning. Compute falls through. A deep-pocketed incumbent makes your mission its top priority. A breakthrough changes the opportunity and the capital required overnight. These are stomach-churning moments, and early belief carries a responsibility to help make the bet work. This is the partnership I love: turning the full ambition into concrete steps. Raising capital, securing compute and strategic partners, recruiting exceptional people, and winning the first customers together. Thinking several moves ahead while solving what’s in front of us. And as breakthroughs open new possibilities, helping founders dream even bigger. Thank you To the founders I’ve worked with: thank you for trusting me and Sequoia to be your first partner and first believer. It’s been the joy of a lifetime, and I can’t wait to keep building together. Thank you to Alfred and Pat for your leadership of Sequoia and your support of my next chapter. To Roelof, for bringing me into Sequoia and believing in me from the beginning. Alfred and Roelof, you have been mentors in investing and in life. So much of what I know, and the investor I’ve become, comes from your guidance and generosity over the years. To Bill, for the close partnership and wealth of knowledge, always wrapped in humility. To Shaun, for showing me how to think deeply and imaginatively. To Michael Moritz, for inspiring 22-year-old me to dream big. To Doug, for teaching me to keep things simple. To Bogie and Jess, for bringing joy to the everyday and perspective on what matters. And to all my Sequoia teammates, for the friendship, camaraderie, and lessons I’ll carry with me. To my family, especially my husband and my brother, who have each started companies of their own: thank you for giving me the courage to take this leap. Let’s get to work!
437
106
75
2,328
376,873
Excited to see Gemini 4 Argon finally released! Over the past few weeks, my team collaborated with another to advance Gemini’s capabilities in game dev. Game dev pushes beyond symbolic intelligence to eval rich spatio-temporal reasoning, a crucial step toward physical AGI.
Gemini 4 Argon is amazing for knowledge work and coding - and it’s super fun to build with! We asked it to create this wizard chess game and it's so cool to see Argon's physics understanding & 3D modelling capabilities, makes me nostalgic for Battlechess on the Amiga - fun times! Argon is rolling out today to Google Ultra and Pro subscribers in the @GeminiApp and Search, and Ultra subscribers in @antigravity! More info & full availability in the blog: blog.google/innovation-and-a…
3
7
110
7,181
Very excited to announce Gemini 4 Argon! It delivers frontier performance across coding, knowledge work and cyber defense capabilities. The model is well-rounded and has great writing quality. The team has been hard at work producing this and we've all enjoyed using it everyday.
Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback. Here’s a look at the benchmarks:
10
37
5
305
28,207
Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback. Here’s a look at the benchmarks:
1,385
2,306
1,225
25,878
3,919,359
Excited to share Gemini 4 Argon! I know many people have been curious about what we’ve been building. We are sharing as soon as possible starting with trusted defenders in our Fairwind Program. We developed Argon with frontier capabilities in coding, knowledge work and cyber security defense. We look forward to making it more widely available as we go through testing.
Replying to @sundarpichai
Importantly Argon has frontier safeguards and we are rolling it out responsibly - it’s with the US gov’t and going to a set of trusted cyber defenders through our Fairwind Program today. We’re going to make it available as soon as we can and as safely as we can. So hold tight, lots more coming, and you’re going to see us iterating rapidly. blog.google/innovation-and-a…
86
104
10
1,799
184,528
I felt a similar despair 4-5 years ago as I was still partially working on robotics. After co-authoring "LLMs are Zero-Shot Reasoners" and "LLMs Can Self-Improve" in 2022, it became clear that digital/symbolic AGI had to precede, and would ultimately accelerate, physical AI. While video models are powerful zero-shot reasoners (video-zero-shot.github.io/), LLMs remain the indispensable core in every aspect of physical AGI development. Over the past year, I've actively collaborated across Gemini, Omni, and generative media to test my hypotheses regarding physical AGI. The pieces are finally coming together. It is clear that frontier LLM labs, alongside the rest of the robotics community, are uniquely equipped to drive this next transition in embodied AI.
I’m sensing deep despair in academics over the past week. Astra, Fable, Muse are zero-shotting benchmarks in robotics & world models. Uneasy pill to swallow, but this is what step jumps in progress looks like.
13
36
4
383
100,498
Beyond RLHF valley
2
767
Mujoco is so back
GPT 6 Astra is capable of orchestrating two robots juggling with one ball at least airborne throughout the exchange This is an actual MuJoCo physics simulation at 1x speed
4
12
93
19,290
We’re also introducing Gemini 3.8 Flash Cyber, our most capable cybersecurity model. It shows frontier-level performance in discovering vulnerabilities and patching them at scale, with Flash-level speed & pricing. That includes achieving 86.2% on the important CyberGym industry benchmark, plus 47.2% on CWE-Bench for patching. We saw a 70%+ success rate in discovering vulnerabilities across 20 programming languages on our internal benchmark.
339
394
88
4,537
417,987
Can Gemini count the number of claps? 👏 Accurately counting rapid movements is a notoriously tricky task for AI. Because static processing ingests video at a fixed 1 FPS by default, split-second movements like a clap easily get missed entirely or get confused with a snap or click. Watch Gemini 3.7 Flash use the new agentic video understanding capability to accurately identify and count every single clap by automatically adapting the processing speed as needed:
21
84
38
847
137,282
Mujoco is back.
Replying to @koraykv
3.7 Flash brings a big jump in agentic performance and coding accuracy. To demonstrate, we set up a 3-agent team to autonomously train a robotics control model from scratch. We hope you like 3.7 Flash, and you can read more here: blog.google/innovation-and-a…
8
6
154
30,259
Today we're launching Gemini 3.7 Flash - our latest workhorse model for coding and agentic workflows, with an introductory price at half the original cost of 3.6 Flash. ⚡️ We have been iterating rapidly with the Flash series, going from 3.5 to 3.7 in just 3 months, making it more helpful across a wide range of tasks: • Software Engineering (DeepSWE v1.1): 37.0% ➔ 65.3% • Web Development (Code Arena Elo): 1506 ➔ 1588 • Enterprise Automation (AutomationBench): 13.4% ➔ 30.4%
118
239
105
2,356
597,964
3 new members have joined my team. Excited to work with them! 1) @rm_rafailov – Co-developed DPO at Stanford. Founding member and Tinker co-creator at Thinking Machine. 2) @arvind_io – Ex-Brain. Co-developed GPT-3 and GPT-4 at OpenAI 3) Jack Ji – RL PhD from Princeton
21
7
3
559
50,752
Rafael joined from outside Google, and Arvind / Jack are internal transfers within GenAI unit of GDM. It's more productive / fun with work with a smaller, flexible team these days.
1
45
4,678
In fact, the first test-time compute I read is the test-time training (TTT). It was coined "dynamic evaluation" by Mikolov and used by Alex Graves / Ilya in old RNN/LSTM days. SoTA compression algorithms like PAQ-8 also had weight updates.
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. much like giving models 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 🥹
4
15
223
29,911
Shane Gu retweeted
Re: Demis, Jeff Dean moves: I think a fair few folks are treating this as bearish for GDM and that is imho a misread. The prospect of reaching AGI and ASI is beginning to look increasingly overdetermined. Being in operational leadership roles to preside over an overdetermined outcome is no longer as high-leverage as being in a leadership role on the next frontier. Early AI/AGI/ASI leads will, over the next year, begin leaving what look like important leadership posts to go place their bets on what they think the most important thing will be.
27
22
7
532
68,070
The "true AGI is the friends you made along the way" slide
Replying to @JeffDean
One more fun slide from our pitch deck.
3
16
2
339
38,404
Shane Gu retweeted
Our self-improving agents optimized the full @vLLM_project inference stack, with up to 16% more throughput and interactivity for @deepseek_ai v4 Pro and @Zai_org GLM 5.2 on B200s (no MTP). Every change was verified and our agents got better and faster at it with each iteration.
12
32
8
257
143,125
Shane Gu retweeted
It was an honor to give one of the invited workshop talks at ICML. The thesis: whoever turns messy real-world outcomes into reliable, scalable reward signals trains models where foundation models alone are stuck. It's also how we pick startups. Talk built with @HenryYin_ Coding agents suddenly started working because code shipped with free verifiers, like compilers and type checkers. We did reinforcement learning on those verifiers, and as we scaled compute, capability compounded. The next unicorn startup are ones that build verifiers for more challenging domains. Periodic Labs grades models with a physics fit against real measurements from its autonomous lab, and the capability ceiling moves. Applied Compute audits LLM judges across tens of thousands of rubric criteria before any RL runs, then trains open models to state of the art on Harvey's legal benchmark. Elorian is building verifiers that give models native visual reasoning capabilities, teaching models to actually see structure, where today's frontier models fail at counting more than ten objects.
Excited to co-organize the 1st "RL from World Feedback" workshop at ICML on July 10, featuring: - @MillionInt : ex VP of RL @ OpenAI - @chelseabfinn : Stanford/PI - Ben Eysenbach: Princeton - @brianzhan1 : Partner @ striker.vc - @robertarail : GDM - Jesse Zhang: UW
4
3
1
64
13,735