(On mat leave.) Cofounded & running @ml_collective. Host of Deep Learning Classics & Trends. Research at Google DeepMind. DEI/DIA Chair of ICLR & NeurIPS.
San Francisco, CA
Joined March 2013
- Tweets 4,856
- Following 1,024
- Followers 47,434
- Likes 6,205
Pinned Tweet
The opportunity gap in AI is more striking than ever. We talk way too much about those receiving $100M or whatever for their jobs, but not enough those asking for <$1k to present their work.
For 3rd year in a row, @ml_collective is raising funds to support @DeepIndaba attendees.
I can never rewatch my own talks — idk, it's just cringe to me. But I do love rereading my own textual expressions from the past. Like, I forgot I was once active-ish on Threads. (No point in sharing the link as I might never be active there again.) And this one is X-pin-worthy
Rosanne Liu retweeted
Join us.
You will work on the most important problem, with an incredibly talent dense founding team.
Time is limited, spend it working on something that changes what is possible.
Rosanne Liu retweeted
I will be giving a talk at DLCT @ml_collective 🎤 in 3.5 hours on our recent work leveraging evolutionary test-time scaling for discovery 👉 ShinkaEvolve 🧬 @SakanaAILabs
Come and say hi 👋 mlcollective.org/dlct/
P.S.: As always, thank you @savvyRL for organizing 🤗
Introducing ShinkaEvolve, an open-source approach to sample-efficient LLM-driven program evolution 🧬
Paper: arxiv.org/abs/2509.19349
Blog: sakana.ai/shinka-evolve/
Code: github.com/SakanaAI/ShinkaEv…
The AI Scientist, Darwin Goedel Machine, and AlphaEvolve have fundamentally shaped my perspective of scientific hypothesis testing as tree search 🌴
While LLMs provide smart recombination operators, effective program mutations remained critical challenges for open-ended and sample-efficient discovery. ShinkaEvolve aims to address this problem and has been battle-tested.
I used to think I became a parent too late in life, but there's a special charm to having lived long enough to think most things are uncool, only to rediscover them as cool again through children's eyes. The best cure for cynicism is seeing your cliché become their new favorite.
Fast R-CNN was 10 years ago. Wow. That was my first implemented network (back then re-implementing papers just to understand was a thing). Remember VGG16? 🥹
Well deserved Test of Time award at ICCV; congrats @inkynumbers !
Congratulations to Ross for winning yet another, Test of Time Award for his high-impact (single author 😲!!!) paper Fast R-CNN! At lunch today, he told us the story behind the paper and his journey and let me tell you, it was FASCINATING. I tried to convince him to share it publicly, but he didn’t agree so I’ll leave it to you all to ask him about it next time you see him! @inkynumbers
It’s amazing to work with someone so incredibly smart yet humble. #ICCV2025
Rosanne Liu retweeted
I built a fun little Mac app called "Outer Loop".
It's a specialized web browser that is aware of SSH. It's good at connecting to web apps running on private servers. Use it for Tensorboard, Jupyter, etc.
See blog post and homepage below.
Rosanne Liu retweeted
Thanks to this wonderful community, I got selected for the Google DeepMind scholarship for a Master's in Mathematical Science (AI for Science) at the African Institute for Mathematical Sciences, South Africa.
Cheers to the next chapter🎉
Rosanne Liu retweeted
I’m recruiting PhD students for 2026! If you are interested in robustness, training dynamics, interpretability for scientific understanding, or the science of LLM analysis you should apply. BU is building a huge LLM analysis/interp group and you’ll be joining at the ground floor.
Life update: I'm starting as faculty at Boston University in 2026! BU has SCHEMES for LM interpretability & analysis, so I couldn't be more pumped to join a burgeoning supergroup w/ @najoungkim @amuuueller. Looking for my first students, so apply and reach out!
The biggest change of coming back to work from a 6-month leave is that I started having nightmares of missing meetings — in those 3am panicky dreams I'd realize I haven't checked my calendar all day and have made a lot of people wait and stare at an empty screen disappointed.
Rosanne Liu retweeted
On today’s MLC-Ng Sunday Specials call, an old-timer returned after a research role at Max Planck l (undergrad at UNILAG); another casually mentioned his new role at Google.
Joining these calls weekly is humbling-seeing so many talented folks doing big things from small places.
Rosanne Liu retweeted
I'm starting a new project.
Working on what I consider to be the most important problem: building thinking machines that adapt and continuously learn.
We have incredibly talent dense founding team + are hiring for engineering, ops, design.
Join us: adaptionlabs.ai/
This thread ("thread" is a mega undersell) will shake you. You should read it well and read it often.
I felt chills reading it while holding a baby in my arms as I realize this new world he's about to figure himself out in will be so different and I can teach him nothing.
Title: Advice for a young investigator in the first and last days of the Anthropocene
Abstract: Within just a few years, it is likely that we will create AI systems that outperform the best humans on all intellectual tasks. This will have implications for your research and career! I will give practical advice, and concrete criteria to consider, when choosing research projects, and making professional decisions, in these last few years before AGI.
This is my current go-to academic talk. It's mostly targeted at early career scientists. It gets diverse and strong reactions. Let's try it here. Posting slides with speaker notes...
--
The title is a play on a very opinionated and pragmatic book by the nobel prize winner ramon y cajal, who is one of the founders of modern neuroscience.
To get you in the right mindset, on the right we have a plot of GDP vs time.
That is you, standing precariously on the top of that curve.
You are thinking to yourself -- I live in a pretty normal world.
Some things are going to change, but the future is going to look mostly like a linear extrapolation of the present.
And the plot should suggest that this may not be the right perspective on the future.
This plot by the way looks surprisingly similar even if you plot it on a log scale. We didn't stabilize on our current rate of growth until around 1950.
A night just like any others. I was pumping milk, my baby awake but quiet. Then shaking happens — baby still quiet, my partner still asleep. I was the only one witnessing it. Night continues, and I gained more appreciation for simply existing. Hope everyone is safe.
Being a parent is to grasp how hard it is to simply exist, is to say "poor baby" a thousand times a day, is to have empathy for all living beings, is to be awed at how the human race has progressed this far, with all the unseen and untold parental assistance behind every step.
Today we say goodbye to @DeepIndaba after six inspiring days in Kigali rich with keynotes, tutorials, workshops, mentorship circles, and insightful posters that kept us learning non-stop. Some of us were only able to make it down to #DLI2025 because of your generous support.
The opportunity gap in AI is more striking than ever. We talk way too much about those receiving $100M or whatever for their jobs, but not enough those asking for <$1k to present their work.
For 3rd year in a row, @ml_collective is raising funds to support @DeepIndaba attendees.