Our Optuna paper has reached over 10,000 citations! A huge thank you to the amazing co-authors, contributors, and community! 🌀🚀

gpt5-pro is superhuman at literature search: it just solved Erdos Problem #339 (listed as open in the official database erdosproblems.com/forum/thread/3…) by realizing that it had actually been solved 20 years ago h/t Mark Sellke for pointing this out to me!

The Applied AI group at UChicago Booth is hiring this year! This was the only non-CS faculty position I applied to during my search, and it turned out to be an incredible fit: more compute than most CS depts I interviewed with, tons of research freedom, and (best of all!) no

Columbia IEOR is hiring for an open-rank faculty position: apply.interfolio.com/175698

🧵 New paper: LLM-ERM: Sample-Efficient Program Learning via LLM-Guided Search arxiv.org/abs/2510.14331 We use reasoning LLMs to learn tasks like IsPrime from ~200 samples by proposing short programs, making both the learned function *and* the learning process interpretable 🤯

“On a log-log plot, my grandmother fits on a straight line.” -Physicist Fritz Houtermans There's a lot of truth to this. log-log plots are often abused and can be very misleading 1/5

In 1983, a Pixar engineer pulled off one of the craziest C hacks ever. He merged a switch and a do-while loop just to make code faster. It looked illegal. It wasn’t. It’s called Duff’s Device.

It is silly to think that economics is a science. It is also silly to think that economics is only worth doing if it is a science

Extremely proud of my sister 😊

About ten years ago I became familiar with the tcs trend of saying “our work solved X year old open problem.” What you learn quickly is the phrase doesn’t imply significance, and many times it means that no one knew about it or cared enough to solve it.

Is it okay if I call this article bullshit?! They didn’t ask me about this; I’m a scientist too, you know😅 I love it here on X, but I find Bluesky boring & depressing. Also, I dislike it when they strongly imply that all scientists must share the same ideology or think/act same!

the training loss seems to be converging

Pretty much every piece of coverage on this paper I've seen is a misrepresentation of what it says. For starters, this is a measure-theoretic statement. The probability of something being 0 does NOT mean it never happens. The main result uses a theorem on analytic functions so-

(hot take??) this paper shouldn’t update you much and I’m very confused why it’s receiving so much attention on twitter

Do you know what can be even more fun? not to go to work on Saturday!

I left Uber this month after finishing a one-year leave from Stanford. What a fruitful year it for me personally. In short I'd suggest any academic in OR to take some time and work full-time in a product-adjacent industry position. A few thoughts: I've done a lot of consulting

🤔 Finished my ICLR 2026 review batch. A recurring theme emerged: many submissions largely followed a familiar recipe: collect a tailored dataset, train a domain-specific model, or introduce yet another benchmark. While technically sound, the majority offered limited conceptual

Fun fact: I got into physics in middle school in part because i was thrown out of the school’s seventh grade math Olympiad team for being too bad at math (the coach was genuinely looking out for me to make sure I (and he) didn’t waste time). I’m still struggling with math, but I

Creativity takes many forms -some innate, others acquired through context. But asking the right questions is almost exclusively a learned skill, requiring time, practice, and study. Most creative people work on questions posed by others; few are bold enough to ask their own.