
Dev Engineer @SlackHQ 🤝 @Salesforce , @USC Trojan, Prev: @PayPal @Braintree @Shutterfly, #AngelHack SF'16 Winner, Founder-Jarvis, @mvpaward @sewausa @gitpro
Joined April 2009
- Tweets23.9K
- Following1.7K
- Followers1.6K
- Likes3K
Replying to @adamfeuerstein
Worth pinning the labeled setting: RASONQUE (daraxonrasib) is FDA-approved for adults with metastatic pancreatic adenocarcinoma who have had at least one prior systemic therapy, or who aren't candidates for multiagent chemo. Once-daily oral. Approved Aug 26. Company also flagged that the early revenue mix includes EAP transitions and initial channel stocking, so the first partial quarter isn't a clean run-rate. $RVMD
Vasu J retweeted
🪔 This Diwali, help put food on more tables!
Sewa International USA is participating in Sewa Diwali, a nationwide food drive with 800+ organizations. Join us in 21+ cities across the USA! Find a contact near you.
ow.ly/hObB5103PHc
#SewaDiwali #SewaUSA #Foodsecurity
Vasu J retweeted
Interesting Amazon putting this out aboutamazon.com/news/company…
Vasu J retweeted
Last month, we demonstrated high-rate alchemy: turning mercury into real, stable gold, beating CERN’s record by a factor of 10.
Last week, we enriched isotopes in the world’s most powerful plasma centrifuge, beating the Soviet record that stood for 40 years.
Now, we’re building a fusion reactor in San Francisco to achieve record power density.
I wasn't super comfortable with AI Human video trainings, this takes it far beyond ...
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
Readers added context they thought people might want to know
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard.
cellcog.ai/blog/tavus-gri…
research.nvidia.com/labs/amri/proj…
tech-ish.com/2026/10/02/tav…
十年前,一个MIT大一新生,在冰箱里装了一个摄像头。
他怀疑室友在偷吃他的东西。
计划很工程师:拍下来,写个AI,当场抓获。
小偷没抓到。
他先被自己拍下的海量视频淹没了。
那一刻他想明白了一件事:
AI 缺的不是算法,是数据。
要做成它,得有现在一百万倍的数据。而这堵墙,每个做AI的人都会撞上。
这个新生叫 Alexandr Wang。
他出生在新墨西哥州的洛斯阿拉莫斯,美国第一颗原子弹诞生的小镇。
父母是中国移民,都是镇上国家实验室的物理学家。
六年级,他第一次参加数学竞赛,目标是赢一趟迪士尼之旅。
他赢了。
后来一路打进美国数学奥赛集训营、美国物理国家队。
17岁,他一个人去了硅谷,在旧金山四处借住沙发,挨个找CEO自荐。
简历几乎是空的,他只说了一句:我热爱写代码,我会跑得飞快。
Quora 创始人 Adam D'Angelo 收下了他。
在 Quora 干了两年,他才去读 MIT——
然后,就有了冰箱里那个摄像头。
2016年夏天,19岁的他和搭档 Lucy Guo 进了 Y Combinator。
两人在一间泳池小屋里办公,晚上睡气垫床。
他跟父母说,这只是个暑期项目。
然后,再也没回学校。
公司叫 Scale AI,干的是当时AI圈最不性感的活:数据标注。
别人在研究改变世界的模型。
他在雇人给图片画框,告诉自动驾驶汽车:这是个行人,别撞。
然后,风口一个接一个砸了下来。
先是自动驾驶,Waymo、Cruise 都成了客户。
再是美国空军和陆军。
2021年,Scale 估值 73亿美元。
24岁的他,成了全球最年轻的白手起家亿万富翁。
一年多后,ChatGPT 横空出世,大模型训练离不开人类反馈数据。
OpenAI、微软,都在用 Scale 的数据。
2025年4月,Meta 发布 Llama 4,口碑崩了,还陷入刷榜争议。
据杨立昆后来透露,扎克伯格非常恼火,对参与项目的所有人都失去了信心。
两个月后,扎克伯格出手:
143亿美元,买下 Scale 49% 的股份。
附加条件:Wang 本人必须来 Meta。
28岁的他,成了 Meta 史上第一位首席AI官,执掌新成立的超级智能实验室。
扎克伯格在内部信里,称他是"这一代最令人印象深刻的创始人"。
Wang 在给 Scale 员工的告别信里写:这一次的代价,是我的离开。
这个决定,难在哪里?
Meta 的AI阵营里,本来坐着一位图灵奖得主,杨立昆,深度学习三巨头之一,在 Meta 待了整整12年。
重组之后,他要向这个大学没读完的28岁年轻人汇报。
2025年11月,杨立昆宣布离开,出去创业。
后来接受《金融时报》采访,他这样评价这位年轻的上司:
"没经验。"
他承认对方学得很快,但说他根本不懂研究是怎么做出来的。
质疑声从来没停过。
媒体爆出,他在会上和 Meta 的老将们意见不合。
2026年3月,Meta 另起了一个AI工程团队,不归他管。
有媒体直接起标题:扎克伯格可能已经"受够了"他。
Meta 出面辟谣,扎克伯格在 Threads 上晒出一张两人挽着胳膊的合影。
一个月后,超级智能实验室交出第一份答卷:Muse Spark。九个月,把 Meta 的整套AI技术栈推倒重来。
有评论说,Meta 总算重新回到了牌桌上。
2026年9月8日,Meta 发布个人AI智能体 Muse。
订行程、回邮件、填表格,甚至替你讨价还价。
它冲上了苹果 App Store 免费榜第一。
第三方数据显示,它上线头12天的日活,是 ChatGPT 当年同期的将近3倍。
9月23日的 Meta Connect 大会上,Wang 宣布:Walmart、Best Buy、Gap、Sephora、Wayfair,都在接入 Muse。
十年前,他雇人给图片画框。
现在,他的AI开始替人下单。
杨立昆说他没经验。
这话没说错。
但 Wang 对此,有另一套说法:
"天真,是有巨大溢价的。"
不知道一件事有多难,才敢一头扎进去。
专家脑子里装着一张"此路不通"的清单。
创始人手里没有这张清单,所以会先去把路走一遍。
扎克伯格押的,就是这份溢价。
杨立昆也没有输。
他带着"大模型走不到AGI"的判断,拿到了超过 10亿美元融资,去做自己坚信的世界模型。
这从来不是谁对谁错的故事,是两条路线,各自押上了全部筹码。
赌局还没结束。
但扎克伯格这一注,第一张牌已经翻开了。
是张好牌。
Vasu J retweeted
The State of California has declared a state of emergency to help California prepare for El Niño. Click through the slides to learn more about what El Niño is, how it will impact us, and how to get ready.
Visit bit.ly/SJStormPrep for tips and resources.
Vasu J retweeted
my WHOOP subscription expired so i had codex reverse engineer it and build my own app
- telemetry goes to my prometheus server
- grafana and the app visualize it
- hermes monitors metrics and plans workouts
next wanna build my own Zwift and Strava
the era of personal software is here
Vasu J retweeted
Cybercab picking me up at my hotel to take me to breakfast. You’ll feel like a VIP getting into this car!
Vasu J retweeted
OpenAI connected ChatGPT to Epic today. Read-only. It can pull notes, labs, meds and specialist documentation, and it cannot write a single character back into the chart.
That constraint is correct and it is also the ceiling. Summarizing a chart was never the hard part of my day. I can skim a chart. What I cannot do in a fifteen minute visit is hold this patient's histology, line of therapy, and organ function against the eligibility criteria of forty open trials, then check whether the drug I want is covered.
Which is why the quieter half of the announcement is the more interesting one. Alongside Epic they shipped a plugin into nine public sources: ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, PubMed.
Chart plus eligibility plus coverage, in one place, is a genuinely different tool than a summarizer. It is the cross reference that never happens because no one has forty minutes per patient to do it by hand.
UCSF is the pilot. Epic sits on records for more than 325 million patients, so whatever this becomes, it becomes it at scale.
The open question for me is whether read-only survives contact with clinical demand, or whether it is a staging post.
The most powerful companies on Earth were shaped, in part, by a small man wrapped in a blanket in a Himalayan village.
Steve Jobs went looking for him. Mark Zuckerberg was sent by Jobs to find him. He was already dead before either of them ever heard his name properly.
His name is Neem Karoli Baba.
He was born Lakshman Narayan Sharma, in a village called Akbarpur in Uttar Pradesh, around 1900. Little is recorded of his early life beyond what devotees have passed down. He left home as a teenager to live as a wandering ascetic, then returned to a settled life for a time, before taking up that path again.
Devotees say the name he became known by came from an incident on a train. He was travelling without a ticket near a village called Neeb Karori, and was made to get off. According to the story, the train would not move again until he was allowed back on board.
He eventually settled in the Kumaon hills of what is now Uttarakhand, and in 1962 built an ashram on the bank of the Kosi river, at a place called Kainchi Dham. He lived simply, sitting on a raised cot wrapped in a blanket, and taught almost nothing in the way of formal philosophy. Devotees describe his teaching as 2 words. Love, and service.
In the late 1960s and early 1970s, a handful of young Americans found their way to him. Among them was Richard Alpert, a Harvard psychologist who later became known as Ram Dass and wrote a book about the experience that reached readers across the world.
He died in Vrindavan on 11th September 1973.
The next year, a 19 year old college dropout named Steve Jobs travelled to India specifically hoping to meet him. He arrived at the ashram to find it nearly empty. Neem Karoli Baba had been dead for months.
Jobs never met him. He stayed in India anyway, and the trip is described in his authorised biography as shaping the simplicity and intuition that would later define his approach to building products.
Decades after that, Jobs told a younger entrepreneur to make the same journey. Mark Zuckerberg has said that when Facebook was going through a difficult period, before the News Feed existed, Jobs advised him to visit Kainchi Dham. Zuckerberg went, planned to stay a day, and stayed 2.
He has said the visit helped him think more clearly about what he was actually building.
Julia Roberts has said a photograph of him was part of what drew her toward Hinduism. Larry Brilliant, an epidemiologist who helped eliminate smallpox, was a devotee and later brought Larry Page and Jeffrey Skoll on the same pilgrimage.
Whether the events attributed to him were miracles is not something an outsider can verify, and devotees themselves describe them as beyond ordinary explanation. What can be verified is smaller, and in its own way just as strange.
A man who owned almost nothing, who left no written philosophy behind, is still quietly present in the founding stories of some of the companies that now shape how billions of people communicate.
Vasu J retweeted
Who really works in the U.S.? 🇺🇸
Out of 161.5M total employed workers: US Citizens:132.5M (82%) • Green Card Holders: 12.8M (7.9%) •Undocumented: 8M (5%) •EADs/OPT/TPS: 5.5M (3.4%) •ALL Temp Visas: 2.7M (1.7%)
Active H-1Bs? 600k (0.36%).Only 0.36% of 161.5 million workforce ?
Vasu J retweeted
Replying to @VivekSubbiah @DrChoueiri @montypal @neerajaiims @doctorpemm @chadinabhan @teamoncology @NarjustFlorezMD @weldeiry @Dr_R_Kurzrock @SewantiLimaye
1/ I'm a cardiologist, not an oncologist. But I've watched pancreatic cancer take patients' family members for 25 years, and I never thought I'd write this sentence.The wall just cracked. Here's what happened.
2/ Pancreatic cancer earned its reputation. It hides until it's late, it shrugs off chemotherapy, and second-line options bought survival measured in single-digit months.The engine behind more than 90% of these tumors is a mutation called RAS. For four decades RAS was the definition of "undruggable" — we could name it, image it, curse it. We couldn't touch it.
3/ In the Phase 3 RASolute 302 trial, an oral RAS inhibitor — daraxonrasib — went after that engine directly.Patients who had already failed chemo lived a median of 13.2 months, versus 6.7 on standard treatment. The risk of death dropped about 60%. Survival roughly doubled.When the data hit the ASCO plenary stage, the room gave it a standing ovation. It was published in the New England Journal the same day.
4/ Where things actually stand, honestly:— Not FDA-approved yet. The FDA accepted it for review this summer on a priority pathway; a decision could come in months.
— Already reachable for some through an expanded-access program.
— First-line combination studies are posting response rates near 60%.It's not magic and it's not a cure. It's proof.
5/ Proof that "undruggable" was a temporary word. Proof that a disease written off as universally fatal can be rewritten. Proof that the next chapter is being drafted in labs and clinics right now.For every person who was told there was nothing left: there is something now — and more is coming.
Hold on. Science just kicked the door in.
Vasu J retweeted
🚨🚨🚨⭐️HUGE news & a MAJOR milestone in oncology‼️
👉🏼The FDA just approved the first targeted therapy for metastatic pancreatic cancer a breakthrough in precision oncology and a landmark step in making the BIG KRAS officially druggable ( beyond KRAS K12C) ❗️
$RVMD
👉🏼A new era in the fields battle against a devastating Emperor of all maladies. #precisionmedicine #KRAS
@OncoAlert @oncodaily @DavidHongMD @Aiims1742 @DrShubhamPant @GIcancerDoc @EileenMOReilly
fda.gov/news-events/press-an…


































