Got interested in physics in middle school in China. Tried physics Olympia in high school. Wasn’t able to make it out of my city in China while classmate got international gold medal representing China with a full ride into Harvard. Got Depressed and thought I won’t make it in China. Took English classes and aced TOFEL. Attended UIUC to study engineering so I’d a job one day, same year when @drfeifei became an assistant prof there. Got an algo trading internship by luck because I wrote a paper on wireless communication. Met MIT interns who told me they didn’t have to pay tuition. Inspired. Got into MIT and studied probability and info theory. Became prof at Stanford. Married. Amazing kids. Pretty happy. So yeah, Thank you, physics!

Today, I became a US citizen.

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 think I did okay considering surviving a pretty theoretical phd from MIT. Lessons: - don’t give up your dreams - don’t fuck with kids who survived highschools in China

I don't get it. what's the deal with europeans and AC? was in Paris one summer and freaking melted me

I dunno that looks like some cheeky satire. Lizzo seems to have a pretty good sense of humor.

at least he didn't have Ling Ling's scholar page on the other monitor

Never say never

For real. Be patient is the key and keep trying/building

I think he worked for some hedge fund then moved to Japan with wife and kids and did his own thing. Didn’t do jack physics after but boss move

Have you tried emailing their front desk secretary? @levelsio Half serious question During my postdoc in Paris i. 2015 I had a checking account with Societe General Paris. I forgot my password one day and tried to change it. I was expecting some kind of text message verification, but instead they said I’ll have to wait for a confirmation code in PHYSICAL MAIL?!! Instead I just emailed the lady they assigned me from my Gmail (en francais ofc) No verification, no question, no password, she just resetted the password for me. This is why things get fucked up when you have red tapes. People just get fed up and go with insane workarounds

I have a question for the Operations Research & Management Science researchers out there: What are some of the high-quality, publicly accessible datasets available for OR research, such as supply chains, logistics, pricing, etc.? Think of the operations equivalent of ImageNet for bench-marking / testing new research and algorithms. Some examples I can think of are the Microsoft MIND Data Set for recommendation systems (msnews.github.io/) and the Technion SEE Lab for data on healthcare and call center operations (seelab.net.technion.ac.il/). MSOM has also run data science challenges in the past, but these datasets are no longer accessible after the contests ended. Are there other good ones out there?

I’ve thus far been sympathetic to Epic‘s struggle, but this is just bullshit what you are saying. They are complying with the law.

nope, just an Asian pig farm

They are literally getting married. You do this in China only at your wedding

Yep, you know. I placed above the top .1% of my province in the college entrance exam (300/400,000). That wasn't good enough for Tsinghua or Peking U. It's hard for people here to comprehend.

I must admit I’m jealous. “As a Chinese” just doesn’t hit the same

Sorry to hear bro. Love this app. This sounds like a hit job from the middle management

Sometime ago i was contacted by some university admins about some procedural things that were totally unnecessarily complicated. I got spammed so much that I put them into a spam folder and forgot about it. This week while fishing for a coupon I accidentally discovered that all their emails went to spam the last year. But nothing happened and all was fine this whole time

It was fantastic to host @rauchg for a guest lecture in our class on the future of @vercel and the agentic web! What a blast and thank you for coming Guillermo!

This is true but it’s also very hard to get past the lottery due to abuse by large consulting slop firms. Many of my friends couldn’t stay because they ran out of clock. What do you think will be a better way to filter out the abusers?

Even a technical paper should be readable in the first 5 pages. I mean literally if it is read out loud in English, word by word, it should sound intelligible and engaging. people often thought this was some kind of metaphor, because surely we are serious people and use serious math. I actually mean it literally. read the passage symbol by symbol and you will see what I mean. Here’s an example from my paper. That line will be read as “ The fact that e sub u tilde of tilde l of u of tilde tau l is less than infinity for all tilde u is readily verifiable from equation seventy six” Stuff like this basically forces readers to stop reading but start symbolically pattern matching for comprehension. Keeping readers enjoy actually reading will make your work more appealing and likely better received, all else being equal.

Ouch, you need new friends bro. But I get what you are saying. A non-Asian friend once described having dinner with my Chinese friends as "an entire evening about finding loopholes"

As the H1B situation unfolds, let me just reiterate what I wrote in a post yesterday: I urge the administration to prioritize H1B visas for students educated in US Universities. I see many dignitaries in tech and in US universities blast the new H1B fees. Most of these people are US-born and never had to deal with how fucked up H1B has been. So, please, understand it first or shut up. For starters, let's understand how the H1B slots were allocated prior to the Trump EO this weekend. First, national labs and universities hires (profs, researchers) are in what's called non-cap exempt category, meaning that people like me would get H1B as soon as we are sponsored by a university. The capped H1B contains 85,000 annual slots. Note that this is 85k slots for the entire United States per year (imo not that many). Amongst them, 20,000 slots are reserved for people holding advanced post-graduate degrees (masters & phd) by INA 214(g)(5)(C). If the goal is to only retain the most technically advanced/talented labor pool, then I'd argue that 20k per year for the entire country is very small, and also the cutoff of a masters degree is woefully arbitrary: why should we prioritize someone with a random masters degree over an undergraduate in a critical industry? Now the problematic part: the other 65k slots are up for grabs for anyone inside OR outside of the US. Foreign firms can (and do) submit a large number of applications for this quota and drown out applicants domestically. While domestic students have some buffer to try the lottery more than once, they are ultimately no match of large foreign consulting firms like Infosys. When these firms send foreign workers to the US, they take up slots for US-educated immigrants, and to make matters worse, there's widespread evidence that the companies themselves pocket a significant portion of the paychecks and rob their own employees (epi.org/publication/new-evid…). In other words, these slots do not fatten the wallets of neither American nor even foreign workers, but enrich their American and foreign EMPLOYERS. Here to me is the hardest part: these dynamics push out students who came to the US to study in the universities, excelled, landed top jobs and wanted to build a family here for the long haul. They ended up having to compete with the foreign consultancies for those few H1B slots and many people ended up having to move back to India, China or to Canada. Adding a 100k fee may or may not end up being the right policy, but something has to be done to this allocation mechanism. I think the 100k fee being a one-time thing is right on, because this rewards people to use the H1B visa for long-term employment and makes short-term consulting engagements economically unattractive. But more needs to be done. For starters, here are my recommendations: 1. Significantly expand the 20k slots allocated to US-educated advanced degree holders to 50k+. Expand the preferential treatment to undergraduate degree holders with excellent academic records. Many shakers of AI today are young people without masters/PhD degrees. They should have a chance to contribute without having to study for a masters degree just to get the Visa. 2. Waive the $100k sponsorship fees if the applicant is US educated. 3. Significantly crack down, via sponsorship fees and lawsuits, on the abuse of the work visa program from overseas consultant firms and domestic companies that are complicit in crowding out the opportunities of domestically educated college graduates. Shrink the quota for overseas applicants. Overall, the goal should be to use this program to filter out the best and the brightest that the US would want to entice and keep in this country. The H1B visa program is a cornerstone on how legal immigration is mediated, and it should not be abused for importing temporary labor and profiteering for foreign consulting firms.

Holy shit this got 1M views?! Physics dropouts broke AI twitter this week

@naval how do you find these bangers if you followed nobody?

real-time footage of software engineers waiting for GPT 5 to drop

Was talking with an immigrant Uber driver who happened to be from near my hometown: he’s so concerned of the lack of substance and rigor of the California public school education, coupled with radical polarization, that he’s now planning to send his (American) kids back to China to be educated. Let that sink in.

Amazing! We built this at @UofIllinois 17 years ago because it was very hard for undergrads to get research experience. Faculty tends to believe there’s low ROI out of underclassmen so most universities barring MIT and the like essentially have zero research opportunities for lower grade undergrads ursa.grainger.illinois.edu/

Honestly, a killer feature would be if these arms help fend off home invaders if someone were to break in when I sleep.

holy smoke how does this guy come up w so many bangers??

About 10 years ago I reached out to Prof Herman Chernoff (yes, *the* Chernoff) as a fresh postdoc asking to meet after reading and loving his design of experiments paper (the man basically invented best arm identification and that was ignored for 50 years). He picked me up from the light rail station outside of Boston in his car and drove me to his nursing home where we chatted and I also met his wife. He shared with me so many cool stories, about how he was a student of Abraham Wald, sharing an office with Kenneth Arrow, hired to build out MIT stats before being kicked out (ironically, that was a time when MIT was trying to build out big data again). … after all that, he drove me back to the train station. What a cool dude! Wild to think of it now. May he continue to enjoy great health!

That’s a wrap! This week officially concludes the first offering of OIT 551, AI and Data Science: Strategy, Management and Entrepreneurship @StanfordGSB (aistanford.org), the first Stanford class focusing on AI and Data Science strategy and management. While the AI world has seen crazy changes in the last few months, the impetus that motivated Luis and I to build this class is as relevant as ever: there remains a big gap between the world of AI / Data Science methodologies and that of management practices. On the one hand, AI tech is advancing at light speed, while on the other hand our understanding of managerial best practices, that of how to set out an effective AI strategies, manage data science teams, control technical risks in ML-heavy products and evaluate AI startups as an investor, are just starting to catch up. I want to thank all of our guest speakers whose insight made the class grounded and extra special! - AI MVP: David Golan (VizAI), Eyal Gura @eyalg (Zebra) - An investor’s perspective: Martin Casado @martin_casado (Andreessen Horowitz), James Currier @JamesCurrier (nfx), Shaquille Vayda (Lux Capital) - Building experimentation capabilities: Su Wang (Lyft) - Cross-functional Data Science teams: Niva Ran @milkandrice (Apple) - AI and Data Science in 2023: @guido_imbens (GSB), and Henrique Ponde (OpenAI) Looking forward to teaching this course again next year, and one can only imagine the state of the filed by then.

Maybe it would help your cause if you are less retarded and learn to read. You ain’t accelerating shit.

A piece about my research and that of my field. It was awesome to work with the @StanfordGSB and the amazing writer @kt_gilb on this - thank you! gsb.stanford.edu/faculty-res…

Bro got upgraded from United airlines into the Singapore air private suits

Here's the transcript (not sure how to add subtitles here): Hey guys, wanted to record this to just share that today I became a U.S. citizen. I just finished my ceremony. First of all, I'm just very grateful. I'm very grateful for everybody, the people, the teachers I had back in China and the career, that education I had here in the United States. So today I became a citizen here. I just want to tell a bit of my story. My family, I grew up in China. My family were half peasants and half merchants in Ningbo, Jiangsu, Shandong. I grew up, I stayed in China and when I was 18, family burned all their savings to send me here. Went to University of Illinois as an undergraduate, then to MIT. Into Stanford as a faculty and now I'm a tenured professor at Stanford. I probably haven't felt so acutely about becoming a citizen here until recent years. Because I feel something's changing in this country. When I came, things are... Things felt very optimistic. People are confident, they are constantly building. And today I walk around Palo Alto. I feel people are shameful, they feel guilty. They collapse inside, they stop dreaming. Recently I had an incident where I went to a doctor and he was very sweet. He, he, he spoke Chinese to me. So I asked, why are we speaking Chinese? And he said, oh, you know, I just feel I shouldn't be this... You're going to run dumb privileged American who only speak English. And something about that really shocked me. This is in Palo Alto in California. And I told him that: It's so sweet that you're learning this language, but you are not dumb. Your country is not dumb. And you shouldn't be shameful. And you certainly should use your privilege for something good. Build something, be proud of it, take the responsibility of it, and don't crawl inside. And don't apologize for that. Do something about it. And that's the overwhelming feeling I have today. I love the quote from MLK who says, something like, darkness cannot drive out darkness. Only lightness can drive out darkness. And hatred cannot drive out hatred, and only love can drive out hatred. And that's the energy I missed. So I just want to conclude by saying that, you know, I'm excited to build. I'm excited for the dreams that this country has afforded me. I cannot imagine my career in any other country, as an immigrant. Like JFK said, ask not what your country can do for you. And ask what you can do for your country. And like an old premier of China, Zhou Enlai, said, 为人民服务 (Wei Ren Min Fu Wu), do something for your people. And that's my ask for you guys, you Americans, and for people like me who just became Americans and for people who are going to be here in the future. Think about what you can do, build. Do something good.

What's the best experiment or A/B test to run in a queuing system? Check out our new paper on experimentation under congestion. By leveraging the underlying Markov chain, we can derive exact variance estimates for different experimentation designs. Naive A/B tests fail in these settings due to cross-unit interference. arXiv: lnkd.in/gSzWbbtx

Here comes the predictable narrative that ES was blocked by "the Jews". Spend 2 more hours to do your research. MANY of us are against this, and many of us spoke out. No offense to any Jewish parents. Yes, the ES agenda from PAUSD was antisemitic, and the Jewish community was pissed. But this is by far not a "Jewish problem" only. By the way, literally nobody was "shocked" - everyone I know was emailing them. Call up some Chinese / Indian parents before writing your next ES piece.

Personally, I don't really give a fuck about them winning a Nobel. You do have the opportunity to shape a young mind and enable them to be happier. That's worth a million Nobels.

I am in pain, to see the unimaginable atrocities and brutality that have been levied upon the Israeli civilians by the terrorist organization, Hamas. Hundreds of civilians murdered, babies beheaded, women raped, paraded, executed. I am in shock to see the terrorist sympathizers across our universities, in including my own Stanford University, who try to justify the horrific massacre of the Jewish people. I am appalled by the moral cowardice of the Harvard University administration for failing to strongly denounce such statements by their student organizations. Let me be clear. This is a massacre, a genocide. This is terrorism to the most horrific order. The horror evokes for me the memories of the Nanjing Massacre (in my home province) which saw the Japanese Imperial Army murder 300,000 Chinese civilians in the course of one week. What is more, they murdered and raped with zeal, holding contests to see which officer could behead Chinese men at the highest speed. This is what's happening to Israeli citizens, right now, in this day and age. This impacts all mankind. When we do not stand up to terrorism, it comes to us all sooner or later. When we do not stand up to terrorism, tomorrow it will be our sons, our daughters, our mothers, our fathers, our brothers and our sisters. I stand in solidarity with the people of Israel.

Damn that’s rough. You should take some time off and enjoy life a bit now that you’ve had it all sorted out! Btw I’ve gone to Bulgaria with a group of GSB students. Sofia and Plovdiv! Beautiful country

- FWIW this new statement from @NeurIPSConf sounds okayish to me. Some point out that there's no "punishment" for her, but I am not sure what Neurips can do. - Picard's own "apology" on the other hand does not even mention the word "Chinese" and is a weak nothing burger. I don't think she should be canceled over this, but she should either stand by what she said, or get to the apology directly and honestly. - The earlier post by Neurips calling this a "cultural generalization" was weak and missing the point. The problem of her original statement is not about generalizing a narrow fact about a few Chinese to all Chinese scholars, but that it made a poorly contextualized statement about a specific Chinese student to begin with. Calling it a cultural generalization implicitly acknowledges the validity of this latter claim. - imo there can be room to discuss whether the Chinese education system does or does not indeed foster the desired moral character in our students. However, doing so should be done with care and rigor (as one could easily imagine the sensitive nature of the subject). The problem with Picard's statement was in part calling out the moral failings of some Chinese students in a random AI keynote. This perceived lack of respect and care is why many consider her remark "racist".

Shocking but reasonable! TIL the privately held Cargill had a 150B revenue in 2025, against NVDAs 165B.

Number theorist in the department when someone calls statistics "math"

Anyone has worked with online convex optimization (OCO) in an applied setting? Short version of my question: when would you NOT use online gradient descent? And conversely, what are the benefits of using GD even if you know the form of the function? Longer version: suppose you are in a sequential problem where at step t, you are given a convex cost function C_t(), and is asked to find a sequence of actions x_t so as to minimize \sum_t C_t(x_t). The practical constraints are that a) you know C_t in each period, but not future cost functions. But, your estimate of C_t can be quite noisy b) you can't move very fast, i.e., x_t must be somewhat close to x_{t-1}. The online gradient descent (Zinkevich 2003) seems to be a very natural solution to this problem: at each step, simply take a gradient step with respect to the current cost C_t, and set x_t = x_{t-1} - grad(C_t(x_{t-1}))*step_size Is this practically the best thing to do? Are situations where doing this is clearly sub-optimal by an order of magnitude? (say I don't care about constants or even small differences in regret scaling) I'm reading the great Intro to OCO book by @HazanPrinceton but can't seem to find answers this there. (@AdamWierman @alexolshevsky1 I'd imagine you are experts in this :) any pointers appreciated!)

It’s his stepdaughter bruh

Wife: let’s go on a romantic date! The romantic date: 😍😍😍🥰🥰🥰🥰

What's the problem? Erythritol does not spike your glucose so it's practically not sugar metabolically speaking

I've heard people in Operations Research trying to figure out how to optimally "market" their work to stay relevant and sexy. the field has taken on a myriad of names depending on the ongoing buzz: business analytics, business intelligence, decision systems, big data, data science, AI etc.... But "operations research" remains an accurate description of what the field does: researching how to operate [a business, Uber, a supply chain, a pricing algorithm......] optimally, intelligently and cost effectively. For example, using causal inference doesn't make you a statistician by default. Using causal inference to operate better is an operations researcher who's resourceful at tool calling. If you are a student of operations research, hold your head high and proudly announce when asked that "I study Operations Research". Just take the inevitable follow-up with pride: "Excuse me, operations What??" Go read about the history of the field and its mythical glory of the linear optimization, of George Danzig's Simplex algorithm that powers every love affair at Tinder and Hinge till this day, of the daring naval logistics during the Great War, of the Convex Optimization engine that helps land the Falcon booster (yes, they used a convex solver in C as far as I heard), of the operations researcher whom we lost to the economists who then won Nobel Prize (did you know Prof. Al Roth published in Math of OR in the 1980s?) .... Just remember that you are the intellectual descendant of dragons. you can't build a field where you change the name every 5 years. you can't build a field when you are not proud of who you are. PS: needless to say I've struggled with this myself and hence the above. So don't be too harsh on yourself if you ever did judge the at time questionable notability of our field in the wider societal conscience. We all have.

We finished the last lecture of AI Strategy! This is the first full-quarter course focusing on AI and Data Science for business at Stanford. You can find the course home page here: aistanford.org, along with a 5-minute summary of the key takeaways: piped.video/HjRtK0JguBY?feature…. Thank you my co-instructor Luis Voloch for starting this journey with me two years ago, first through a half-quarter pilot and now in a full course! Thanks also go to all the GSB/Eng students who joined this journey and the amazing TA Samuel Chang, PhD! To all of our guests - thank you for bringing the diverse perspectives and insights to our students! - ML MVP: Shir Meir Lador @shirmeir86 (Intuit) - Horizontal and Vertical Expansion of AI Products: Eyal Gura (MAPS Israel, Zebra Medical), Cameron Andrews @camatsirona (Sirona Medical) - Building Experimentation Capabilities: Su Wang (Lyft) - The Economics of Data: James Currier @JamesCurrier (NFX), Rob Toews @_RobToews (Radical Ventures) - Organizational Design for AI and Data Science: Niva Ran @themilkandrice (Apple), Dylan David Daniels (Haus) - Generative AI and Their Cost Structure: Santosh Raghavan (Groq), Dylan Reid @dreidco (Zetta Venture) - Frontiers of Data Science and AI: Henrique Ponde @HenriquePonde (OpenAI), Shreya Rajpal @ShreyaR (Guardrail AI), Ashwin Paranjape (Samaya AI)

I was very bad at deducing how many chicken and pigs were in the same cage by counting jus the legs.

Yay, meanwhile still wishing someone making AI3 great again!

There’s no way that building is 1500 years old

This month marks my 10th year at Stanford since moving out west. A lot happened, learning how to teach MBA courses, research, mentoring, tenure, kids... Thank you to all of my MBA/PhD students and colleagues for putting up with me and giving me a chance. Lots of bumps along the way for sure, but what a journey it has been.

Hey @mntruell, @cursor_ai could someone help us out here? I'm going to teach with Cursor in our GSB MBA class with 200+students in two weeks but many of our students are having trouble getting approved for your Free 1-year Student Cursor Pro trial. Looks like that your SheerID isn't working (example below). Thanks for the trial & help unblocking this! "Basically once I log into Cursor and step into verifying my student status, Cursor redirects me to SheerID, which asks for a government ID but never accepts it. I've tried multiple times with different pictures of both my student ID and my government (state & passport) IDs, but nothing seems to work. Then at some point you go through so many tries that they just block you from attempting to be verified (see first screenshot). I also reached out to SheerID but they are unable to do an offline verification, and also have just ghosted me, so sadly I have no more paths of recourse left to pursue"

My interview with the if/then podcast is out! Our conversation revolved around how to make AI and data-driven decision making work for a business need and why emotions often get into the ways of clear, first-principle thinking. Also a bunch of stories from the trenches. Check it out! Podcast: piped.video/watch?v=eVwHF2wR… Do you agree? What do you think? Leave a comment down below! 👇👇 #podcast #interview #ai #datasicence #datadrivendecisionmaking

It is a disgrace that Harvard's leadership has failed to firmly and unequivocally condemn the atrocities of Hamas. Thank you for doing this. A similar effort is underway here at Stanford.

This is a not simple issue and it’s obviously up to debate. But you chose to go with a one-liner in your original post presumably to incite outrage. That’s what I, and many others, are responding to. Why didn’t you list this ”question” on your original post to kindle a proper discussion?

Took me a while to realize why Chammath and Fridberg went bald

Probably the most dramatic conference dinner venue I’ve ever been to, wow 🤩 #APSNancy #APS

Our paper on Nonstationary Bandit Learning via Predictive Sampling will appear in @aistats_conf 2023! Main idea is to shape exploration by reasoning about the future value of information. arxiv.org/abs/2205.01970 #AISTATS

Thanks Shaun. This takes courage.

Happening now at #SIOP23Seattle SIAM Conference on Optimization! Check out my student Yueyang Liu presenting her work at 6pm on a definition of non-stationary in multi-armed bandits, and why it matters for designing bandit agents/algorithms! talk: meetings.siam.org/sess/dsp_t… paper: arxiv.org/pdf/2302.12202.pdf @TheSIAMNews

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 over the years where I designed and helped launch high profile AI/data products. But working full-time was an entirely different experience compared to that. The immersion is not something you can get from some research collaboration, or even paid consulting gigs. You learn the world out there is big, that people actually change jobs, and that you can actually survive, and even thrive, outside of a university. This makes you realize a) that you have a conscious choice to make (what to work on, where to work), and b) many choices you made in your academic life may not have been as conscious as you thought. And that's okay. You learn that it's all about the people. If you want to excel in a large company, proactively identify / work with / learn from the best of the best is necessary. You learn that people in industry don't get bogged down by the kinds of bullshit that you've been so used to in academia that you don't even feel it anymore. But magically they do have to deal with other kinds of bullshit that you didn't know existed. You also also learn why these hurdles exist, when/when not they were partially justified. You develop empathy. Thanks to all of you that I had the fortune to work with (too long to list, but y'all know who you are)! LFG

Huge congrats to my student Yueyang Liu (co-advised with Ben Van Roy) for joining the Jones Graduate School of Business @RiceUniversity as an incoming Assistant Professor of Operations Management in 2024! Well done!

It's tele operated, no?

I remember watching American tv shows as a kid in China and saw kids wear outside shoes IN BED. The horror!

I thought about your post a little more. I think at the end of the day, you can't control what other people call you, diversity hire or not, them being sexist or not, the system being fair or not. You ultimately have the choice, of what to study, where to work, even though you may feel you don't, especially with all the sunk costs that come with academia. But you in fact do. With the recognition of that choice, comes power. Setting the boundary sets you free. What I'm trying to say is that I sense a feeling of bitterness in your post(s), and I think this is very relatable. I wish you recognize how powerful you truly are, and be nourished by that power going forward, cometh what may.

Oh man it’s been years. I miss that place

When I mention “Jensen’s inequality” in my class (when we teach the flaws of the averages) some people just assume it was invented by Jensen Huang lol

I've had a lot of academics (pre and post-tenure profs in top universities) reaching out to me over the years to ask about my experience working with the industry (and getting paid). A few common themes:

Thanks. Trying. Work is often an escape for deeper wounds

Our paper on private convex optimization has been accepted to the IEEE Transactions on Information Theory! (Fun fact, this is my first IEEE paper despite having a PhD in electrical engineering and computer science! 😱) lnkd.in/gJcxiiB7

Interesting perspectives. I have personally benefited a lot from having had good therapists in talk therapy. They really helped me get through some tough periods in life, and when it got much better, the therapy simply stopped. I count my self as lucky to have found great therapists. But consistent with what's being said in this video, a shared trait of these therapists was that they did not try to convince there was something wrong with me, but rather it was "you are fine, things are tough, and let me help you get through this". Another anecdote. I went through a rough patch during my teenage years. Parents took me to some random psychiatrist (this is was the late 90s in China), who after talking to me for 5 min slapped a clinical diagnosis and prescribed me meds. I took them, but it only made it worse. Desperate, my parents managed to find a renowned expert in my hometown for a second opinion. This person was the head of the department at the local psychiatric hospital. He saw me, asked several questions. Then summoned my parents, and in my presence, told them: your son is fine, nothing wrong. Go home and stop jumping off the wall every time he feels something. Two weeks later, I got much better. I'm forever grateful for this man.

Congrats dude! Well deserved

It's official! I received the promotion to tenure at Stanford University this week. What a journey it has been since coming out here in almost 7 years ago. Many thanks go to my advisor, students, collaborators and colleagues through the thick and thin! Onto the next leg!

Wow this is honestly impressive. Did they fine tune the model on unhinged X hustle bros??!!

And you need to check into a mental hospital

You don’t write bangers. Bangers write you.

Three years ago I met a data scientist who said there was a team at Rivian that worked on vehicle maintenance prediction. I knew it was doomed then.

So I need to write a research statement for an upcoming promotion and I'm really struggling and procrastinating hardcore :( So to unblock I figured I will do a stream on here and Twitch like those gamers and AI bros like this dude @jsuarez! If you want to come see my write boring Latex and awkwardly brag about the entirety of my research portfolio, come hang out! First stream will be tonight at 9pm PST (Sep 13). Link will be posted on my account here

Shana Tova and be easy on her

🙏🙏

I’m not from your field but from the look of it you’re incredibly accomplished!! I do feel academia has this magical power to pull you in and lose perspective. FWIW, I spent a year in the industry and felt there was a huge world out there to explore.

Some tips for new students (and professors)

Great to see the familiar & new faces at another INFORMS! Yueyang Liu (Stanford MS&E yuey7x.github.io) is on the job market and will be talking about non-stationary bandit learning on Monday at 2:15-2:30PM, in CC-North 123. Yours truly will be talking about better ways to analyze switchback experiments under stochastic congestion, Monday at 8am in CC-North 121A.

I’d guess the same dude

Wrapping up the first two weeks of our new class I'm teaching with Luis Voloch on AI and Data Science Strategy here at Stanford University Graduate School of Business. It's been quite a journey putting this class. We are trying to answer the question: what makes the strategy, design and management of AI and DS-driven products and businesses so different. In particular, we see how conventional wisdom around software, SaaS, MVP and organizational design can fail when trying to build a sustainable AI business... We don't have all of the answers yet, but more than excited to explore this new territory with our students! Class page: sites.google.com/view/oit-55…

"Very interesting...."

you posted the wrong guy

Check out our new paper on agentic AI for energy operations! Shoutout to the coauthors Cong Chen (postdoc, incoming professor @thayerschool) and @OmerKaraduman34. Paper: arxiv.org/abs/2506.12664 1/n👇

Check out the newsletter just launched by my lovely wife Niva Ran, on her journey of balancing motherhood with developing a career in STEM #career #stemmom #womanintech #womaninstem milkandrice.substack.com/

Now? Where the fuck have they been?

Disappointing to see politicizing of the classroom of 2-3 year-old children. The center is managed by @ICRIchild icrichild.org/our-team

You seem to have history of anti-China rhetoric. You should publicly apologize

Agreed about the damage to math. But UCLA administration at large also utterly failed their Jewish students. Tao remained silent on this as far as I could tell when hell was breaking loose on campus, as do many STEM people, as if “politics” should just go around them and let them do math.

Our new preprint with Gal Mendelson @TechnionLive studying how adaptive control can "erase" relevant information from the state transitions, and consequently, impact down-stream statistical analysis. We look at the statistical task of detecting service slowdown, but the phenomenon of information erasure due to adaptive control is likely more general. arxiv.org/abs/2401.07305 Work in progress. Welcome feedback + ideas on how to extend this. #statistics #operationsresearch #queueingsystems #datadrivencontrol