@racursion

Visiting Scholar @NYU. Prev: Founding Team & Sr. Director @PrescientDesign ∙ @genentech

Brooklyn, NY
Joined November 2025
Stephen Ra retweeted
Hey there, political scientists and economists. This is a trap.
Political scientists and economists: we're looking for researchers with deep expertise to come work with us for a unique 4-month fellowship at @AnthropicAI. You'll work closely with us to evaluate, improve, and apply Claude's capabilities in your field. It's going to be extremely fun and interesting work, and definitely worth taking a short leave from your university for if you need to! More info here: job-boards.greenhouse.io/ant…
9
9
3
257
33,900
Stephen Ra retweeted
OpenAI’s faster Fourier transform is less a mathematical breakthrough and more a malicious exploit of mathematical traditions. argmin.net/p/faster-fourier-…
67
112
54
1,133
355,300
Stephen Ra retweeted
(1/n) Excited to announce Zatom-2, a generative model trained on ~5M atomistic structures from OMol25 and OMat24! Learning to sample from datasets spanning relaxed and distorted geometries is challenging, and we explore several training recipes and the role of force conditioning
Thrilled to announce Zatom-2, an atomistic generative model spanning molecular, material, and protein science! Special thanks to @MirunaCretu2 for leading this effort. (1/11) 🌐 Project page: zatom-ai.github.io/nucleus
2
11
1
49
2,300
Super Intelligence II Turbo
1
3
129
They will not. This is completely fake. Your company helped Trump snuff out 120 Iranian school girls in Feb & all your CEO said was “it didn’t even violate any of our red lines.” Why dont you comprehend the personhood of actual persons first then I’ll take this bullshit seriously
I continue to think (and fear) that questions around AI personhood are going to be the defining religious conflict of the 21st century. It's going to get extremely ugly.
15
654
4
4,120
60,784
How Cornell, disciplined, punished, expelled and indeed facilitated deportation of students who protested Israeli genocide, while simultaneously providing cover for gang rape in a fraternity is an exemplary tale of our time. Senior leadership of the university should resign.
7
160
2
718
11,360
Stephen Ra retweeted
Philosophies of blind utility maximization are inescapable in today's society. We should learn when they are misapplied. argmin.net/p/the-keynesian-s…
5
2
12
3,293
Another think piece from the collaborationist wing of the professoriate.
The head of the philosophy department at Williams College downplays the pedagogical value of writing: “What if we are mistaken in treating writing as the highest cultivation of thought? The evidence that supports this belief is thin at best. Writing can be useful cognitive scaffolding for memory and self-regulated learning, but the notion that it is the best process for developing thinking skills is a fairly recent idea, traceable to the 1970s writing-across-the-curriculum movement.”
21
23
6
341
44,879
🔥Welcome 🔺Simplex Diffusion Models🔻! Standard discrete diffusion models suffer from "information collapse" because they discard uncertainty at intermediate steps by sampling categorical tokens. (1/5)
16
107
12
844
97,023
it is easier to imagine the end of air travel than non-extinction of humanity
Can Ezra Klein do an interview with the CEO of Airbus and ask them if they “feel pressure” to ship airplanes without testing them first?
2
3
643
Stephen Ra retweeted
Our first protein 🥩 design paper! 🤠 Check out Minkyu's tour de force on T-REX 🦖, an agentic protein design controller for de novo binder design! TLDR we treat protein design as an online allocation problem over heterogeneous tools (Proteina-complexa, Boltzgen, Bindcraft, ✨ etc), use 🤖 agents to reason over what tools + settings to run, designed a granular classification system to manage the agents (e.g. Rescue🛟/Explore🔎/eXploit🚀), that is all orchestrated with a deterministic controller to efficiently brrr the GPUs. 💸 T-REX 🦖 is fully open source: github.com/ml-struct-bio/T-R… Preprint: biorxiv.org/content/10.64898… Please let us know if you have any feedback!
Excited to share T-REX 🦖: Target-adaptive Rescue-Explore-eXploit, an agentic campaign controller for high-throughput de novo protein binder design! T-REX treats protein design as an online agentic allocation problem to leverage diverse tools with efficient GPU usage 🧵
5
45
1
235
20,407
Synthesis is a major bottleneck in drug discovery. AI synthesis planning can help propose better routes and make generative molecular design tractable. With our retroChimera model we have been pushing the boundaries of these models further - out in @Nature today. 1/2
12
38
4
171
17,507
Stephen Ra retweeted
Are organoids useful for Alzheimer’s research? (3.8k words, 17 minutes reading time) owlposting.com/p/are-organoi… the final organoid essay; a case study into how useful organoids have been for one of the hardest-to-address diseases that exist today also: i moved to san francisco!
6
7
4
107
13,694
The Empirical Theory of Scientific Theory, post 1 in my Notes on Recht's Notes on Meehl 1/2
a little project linked below: i'm writing a series of posts following Paul Meehl's final lectures on metatheory of psychology, which @beenwrekt similarly blogged through in 2024. these are notes on Recht's notes on Meehl, and they're about stats, science, and postmodernism 1/2
1
1
2
718
what shall we call this
4
3
25
1,306
One of my favorite passages from "Pilgrim at Tinker Creek" (more or less required reading having grown up on Hollins campus); a passage I often revisit in the wake of my father's and brother's deaths. RIP Annie Dillard
3
363
Today we’re announcing we’ve raised $20M in seed financing led by @QuietCapital , with participation from @GradientVC , @haystackvc and @CompoundVC and other amazing folks to build the verification layer for AI-driven biology. We are also delighted to be welcoming @vqctran who is joining us from @GoogleDeepMind as a co-founder. AI is not going to cure all diseases without a verification substrate - a scalable environment where models can test predictions and learn against living human biology. AI is becoming increasingly capable of generating biological hypotheses, therapeutic candidates and experimental designs. But generating ideas is advancing much faster than our ability to test what those ideas will actually do in human biology. Existing systems force a tradeoff: the most biologically relevant approaches are slow and expensive, while faster in vitro and computational systems often lack the fidelity needed to capture human response. Polyphron’s bet is that tissue is the fastest, cheapest, most parallelizable verification substrate that still contains the biology that matters and we believe the path to biological superintelligence runs through closed-loop interaction with these living and simulated human systems. Thank you to @axios for covering the announcement. Link to our new website in the thread.
47
35
23
274
71,624
Living systems routinely map environmental signals to responses through enzyme-catalyzed reaction networks, but what physical features actually determine their computational capacity? We developed a scalable inverse-design framework that turns chemical reaction networks into function approximators by tuning physical parameters such as transition-state energies and reaction driving forces. Here’s one example: fitting the von Neumann elephant with a chemical reaction network 🐘 We find that network size and local control over reaction driving strongly determine the expressivity of these networks. Paper: arxiv.org/abs/2609.17339 Code: github.com/dayhofflabs/chemi… with @marcotux and Jason Rocks
4
6
1
77
14,418