XOutlier
Download .zipAdd to Chrome

Yuchen Jin

@Yuchenj_UW306k followers

Cooking fun AI systems & products @databricks. Prev: co-founder & CTO @ Hyperbolic, OctoAI (acquired by @nvidia) Apache TVM, PhD @ University of Washington.

Open on X
Top outlier
125×
Best organic post
Peak views
10.2M
Single post
Bookmarks
131k
Across 89 organic posts
Last scan
Oct 16
Community cache
Posts 89 posts
#DatePost↓ OutlierViewsLikesRepostsBookmarks
1Oct 31, '25PewDiePie in 2025: – built a 10×4090 rig – runs Llama 70B, gpt-oss-120B & Qwen 245B locally via vLLM – built a custom web UI (chat, RAG, search, TTS) – ran protein-folding simulati125×10.2M58k2.4k15k
2Apr 27, '25I saw a guy coding today. Tab 1 ChatGPT. Tab 2 Gemini. Tab 3 Claude. Tab 4 Grok. Tab 5 DeepSeek. He asked every AI the same exact question. Patiently waited, then pasted each respo88.5×7.2M124k6.4k23k
3Jun 17, '25Sam Altman says Meta is offering $100M signing bonuses to OpenAI staff. Not $100M annual compensation, just the signing bonus! He clowned Meta: “that’s not how you build a great cu75.7×6.2M12k9274.1k
4Sep 20, '25It's sad to hear President Trump raised the H-1B Visa fee from $1,000 to $100,000. - New grads will struggle to get a job - Startups can't afford global talent - Top minds will go 66.4×5.4M8.2k539919
5Dec 24, '24USCIS just denied my US green card application. claiming my work lacks impact "beyond that of Apple, Inc." -- even though they acknowledge that I am the Apple CTO. WTF? I have NEVE48.3×4M17k1.7k3.1k
6Jun 2, '25>went to a Miami restaurant >12K Google reviews, 4.9 stars >food is mid >bill arrives, way higher than menu prices >"sir, we added 20% service fee + 2% resort tax" &42.4×3.5M43k9101.5k
7Jun 9, '25Ilya Sutskever, in his speech at UToronto 2 days ago: "The day will come when AI will do all the things we can do." "The reason is the brain is a biological computer, so why can't 37.8×3.1M6.8k7133.5k
8Jul 8, '25Sam Altman was asked how he felt about Zuck and Meta poaching OpenAI’s top talent. “Fine... good...” he said. Behind Jony Ive–designed glasses, I couldn’t see his eyes. But I could30.7×2.5M4.3k1321.2k
9Feb 4, '25judging an engineer by age is BS - Linus Torvalds wrote Linux at 21 - Steve Wozniak built Apple I at 25 - Palmer Luckey created Oculus VR at 20 - Vitalik Buterin designed Ethereum 30.1×2.5M27k3.3k2.6k
10Oct 17, '25Jensen Huang: “We (Nvidia) are 100% out of China. We went from 95% market share to 0%. I can’t imagine any policymaker thinking that’s a good idea.” Chinese AI labs like DeepSeek n28.9×2.4M7k8371.5k
11Feb 18, '25Grok 3 might be the best base LLM for real-world physics! Prompt: "write a python script of a ball bouncing inside a spinning tesseract". There is no "thinking" or "big brain" mode25.6×2.1M4.6k575517
12Oct 29, '25SoftBank was the largest shareholder of Nvidia. But they sold their entire 4.9% stake in Nvidia in 2019 for $3.6B. Nivida stock hit $5T today. Masa lost $250B. Jensen Huang making 23.7×1.9M12k9701.7k
13Nov 2, '25Met a Meta AI researcher. He studied Physics in China, came to the US for a PhD in Physics, and then fell in love with AI, despite never having studied computer science. He watched23.5×1.9M8.8k5813.6k
14Nov 14, '25The creator of GPT doesn’t have a PhD. The creator of PyTorch doesn’t have a PhD. The research lead at Cursor dropped out of NEU. You don’t need a PhD or a top school to become a g23.1×1.9M5.8k5301.2k
15Sep 12, '25How money works: 1. OpenAI signs $300B GPU deal with Oracle 2. Larry gains $100B (no GPUs shipped) 3. Larry invests in OpenAI’s $1T round 4. Sam uses $300B to pay Oracle 5. Oracle 22.6×1.8M29k2.2k5.4k
16Sep 13, '25Sam Altman: “I need GPUs, a lot of GPUs.” Larry Ellison: “We used to do databases, but sure, Oracle can build GPU data centers.” Sam: “But we don’t have money, as Elon said.” Larry19.8×1.6M11k6491.4k
17Sep 11, '25“OpenAI signed a $300 Billion GPU data center deal with Oracle.” > Larry Ellison gained $101B in one day > now the richest man alive. > but Sam doesn't have $300B > Lar18.4×1.5M22k1.3k2k
18Jul 14, '25SoftBank was the largest shareholder of Nvidia. but they sold their entire 4.9% stake in Nvidia in 2019 for $3.6B. It's worth $200B today. Jensen Huang made fun of Masa in front of18.4×1.5M6.8k4971k
19Sep 25, '24How it started vs how it’s goin.17.1×1.4M6.8k471683
20Oct 10, '25Interviewed an engineer. I’m 99% sure he was using Cluely. He’d pause for 3 seconds, stare into the void before every answer, then drop a perfectly structured 5-paragraph essay, wi15.1×1.2M20k3612k
21Sep 2, '25People keep saying don’t learn CS and coding, meanwhile Pewdiepie is out here coding on GitHub and building a PC with multiple NVIDIA GeForce RTX GPUs.15×1.2M9.9k3311.6k
22Aug 12, '25“skill issue” 💀13×1.1M19k522721
23Feb 11, '25This is wild - UC Berkeley shows that a tiny 1.5B model beats o1-preview on math by RL! They applied simple RL to Deepseek-R1-Distilled-Qwen-1.5B on 40K math problems, trained at 812.4×1M3k3621.8k
25Jul 30, '25Not a single person from Thinky or Anthropic took Zuck’s $200M–$1B offer. Why? 1. Already rich 2. Mission > Mansion 3. they've seen AGI! 4. hate big corp life 4. upside is highe11.5×938k6.5k2051.5k
26Aug 11, '25If Jensen truly believed AGI was near, Nvidia wouldn't sell a single GPU.11.4×934k7.6k239733
27Sep 15, '25If Jensen truly believed AGI was near, Nvidia wouldn't sell a single GPU. Google hasn’t sold a single TPU chip in 10 years. Just saying.11.3×928k7.1k201924
28May 30, '25Which tech couple are you betting on?11.1×906k6.3k182700
29Sep 20, '25Got flooded with “go back to your country” and “America doesn’t need you” on my H-1B tweet. America is 4.1% of the world. Even if Americans were 10X more likely to be geniuses, 70%10.9×891k6.5k423548
30Dec 31, '24Alec does not have a PhD degree Alec invented GPT and CLIP in Jupyter notebooks you don’t need a PhD degree to be a great researcher and engineer You can just do things!10.8×886k5.5k4891.7k
31Jun 28, '25Zuck poached 4 elite Chinese researchers from OpenAI: - Jiahui Yu: led o3, o4-mini, GPT-4.1 - Hongyu Ren: creator of o3-mini & o1-mini, core contributor to o1 - Shuchao Bi: head of10.7×874k4.6k2611.3k
32Apr 20, '25For the first time, I truly believe AI might replace doctors. I felt dizzy every morning when standing up from bed, and even worse after moving to a hill in SF, the dizziness hit m10.5×864k4.1k2031.2k
33Jul 1, '25Thanks for asking, but can I report to your Chief AI Officer? I have a strong conviction that LLMs can lead to AGI/ASI.10.2×835k4.8k112569
34Jun 10, '25o3-pro is the slowest and most overthinking model. A simple 'Hi' cost me $80. 🥲9.7×795k9.9k167543
35Oct 23, '25Meta laid off 600 people from its Superintelligence Lab today. Many FAIR researchers, including FAIR Research Scientist Director Yuandong Tian, were affected. I think Yann Lecun wi9.6×787k5k290624
36Dec 8, '24It’s so funny that there is a designer at OpenAI who makes $350K - $20M TC that designed this, got approved, and shipped it:9.4×773k6.5k120408
38Sep 22, '25Sam Altman: “I need GPUs!!” Jensen Huang: “You know who has GPUs” Sam: “But I don’t have money” Jensen: “Let's do a strategic partnership.” 9:00am: “Nvidia to invest $100B in OpenA8.8×720k8.4k474812
39Jul 29, '25“Computer science is not science - it was a fake field for people who find electrical engineering or maths too hard.” - Peter Thiel8.7×716k5.6k3111.2k
40Aug 7, '25My OpenAI friends are so hyped rn - not because it’s the night before GPT-5, but because Sam just announced $1.5M bonus for every employee over 2 years. 78% of Nvidia employees are8.7×709k7.5k3061.2k
41Jan 22, '25Wow, DeepSeek-R1-Zero is the least censored LLM I've ever used. Why? It’s trained purely with RL, without supervised fine-tuning. So even if you ask sensitive questions about China8.5×698k3.6k2731.8k
42Jun 4, '25They're just like human programmers.8.5×698k20k6231.1k
43Oct 9, '25How to panic senior engineers with one message.8.4×687k11k298610
44Jun 12, '252013: Google buys Hinton, Ilya & Alex - $44M 2014: Google buys Demis Hassabis & DeepMind - $400M 2025: OpenAI buys Jony Ive - $6.5B Meta buys Alexandr Wang - $15B Let that 8.3×675k5.3k193801
45Nov 4, '25Google: AGI is an energy problem. We’re sending TPUs closer to the sun.8.1×665k4.4k200437
46Jun 14, '25Hot take: You should still learn to code.8.0×657k7.1k411675
47Jan 23, '25China has: - llm: DeepSeek, on par with o1, open-source - robotics: UniTree - drone: DJI - TSMC is next door - can certainly invest $500B in AI What's the USA's moat?7.7×631k6.8k4191.4k
48Oct 25, '25GPT-5: lost 71% in a week. Qwen 3 Max: gained 70% in a week. How is Qwen 3 so good at trading??7.6×624k3.1k1921.1k
49Feb 3, '25The Art of Naming -- OpenAI7.6×618k7.9k312505
50Aug 30, '25Ilya Sutskever: bald Demis Hassabis: bald Noam Shazeer: bald Greg Brockman: bald forget AGI. forget curing cancer. cure baldness now. My hairline is on gradient descent.7.5×611k6.8k313608
51Oct 1, '25Sam Altman’s playbook is clear: When a model becomes usable, OpenAI rushes to turn it into a blockbuster app: ChatGPT – chat app Codex – coding app Sora – video app Raw model power7.3×600k3.2k174863
52Jun 30, '258 out of 12 people on Meta’s Superintelligence team are Asian. 2025 is the year of a̶g̶e̶n̶t̶s̶ asians.7.2×591k4.7k3591.1k
53Oct 18, '25At first, I thought GPT-5 had cracked those math problems on its own. Turns out (as Demis pointed out) GPT-5 just looked up the answers via web search. We really need better peer r7.1×581k2.9k240412
54Apr 7, '25If Meta actually did this for Llama 4 training to maximize benchmark scores, it's fucked.7.0×575k3.2k192926
55Jun 11, '25Should I hit send?7.0×569k3.1k19212
56Nov 11, '25Yann LeCun is finally leaving Meta. An outcome that is inevitable after Zuck spent $15B to acquire Alexandr Wang and made Yann report to him. OpenAI’s LLMs and ChatGPT have made Zu6.8×557k3.8k215592
57Jun 30, '25Was at a dinner in SF when I overheard a big tech engineer say he was just laid off. 5 minutes in, someone at the next table patted his shoulder, and said: “Same. My manager said A6.8×554k3.8k2391.1k
58Jun 29, '25Sorry to hype — but having a few friends at OpenAI makes it hard not to hear how wild their open-source model dropping next month is. It won’t run on a phone, but let’s just say it6.7×548k3k86649
59Sep 3, '25>went to an SF restaurant >"SF Mandate (8%)" on the bill >it's not a tip >it doesn’t go to your server >suggested tips were "15% - 18% - 20%" on top SF has 8% Mandat6.4×525k6.1k114280
60Sep 9, '25There’s a new illness. I call it “LLM Dependency Syndrome.” You can save that $1,200/month by writing simple code: >extract phone numbers: regex >check profanity: blacklist >reform6.2×506k7.1k4042.1k
61Sep 29, '25We were promised AGI, ASI, personal superintelligence. Instead, we will get infinite slop machines that turn us into dopamine-addicted zombies.6.1×498k9.1k842608
62Aug 25, '25Meta’s Superintelligence Lab is already losing researchers? A billion can buy you a house, but not your dream.6.0×492k3.2k123732
63Mar 1, '25holy shit, DeepSeek is able to get 73.7k tokens/s input and 14.8k tokens/s output throughput per H800 node! Their profit margin is 545%, while OpenAI is bleeding money despite char5.9×485k4.7k4371.1k
64Oct 7, '25Google has won 3 Nobel Prizes in just 2 years: >2024: Demis Hassabis (AlphaFold), Geoff Hinton (AI) >2025: Michel & Jone (Quantum Computing) No other tech company matches Google’s 5.6×460k6.3k428715
65Aug 1, '25closed-source AI vs. open-source AI which way western man?5.6×458k6.5k3171.4k
66Aug 6, '25You can just ask for things. Worst case is a no, but it can also be a yes.5.5×454k9.3k133715
67Jul 10, '25You can laugh at xAI folks sleeping in office tents or grinding till 4:20 AM on weekends, but you have to admit: they’re the fastest-moving AI lab out there.5.3×431k5.9k229247
68Nov 6, '25If you ever wonder how Chinese frontier models like Kimi, DeepSeek, and Qwen are trained on far fewer (and nerfed) Nvidia GPUs than US models. In 1969, NASA’s Apollo mission landed5.2×424k3.7k267511
69Jan 26, '25"Tighten export controls on chips" is a loser's attitude. - DeepSeek can already run inference on Huawei Ascend chips - This will only push China to accelerate GPU development and 4.9×401k4.4k379511
70Dec 23, '24I'm more and more confident that tokenization will be gone. Humans don't think in "tokens". Tokens are hardcoded abstractions in LLMs that lead to weird behavior: LLMs can solve Ph4.8×389k3.7k4392.6k
71Jul 11, '25Google is the greatest acquirer of all time: 2005: Android 2006: Youtube 2013: Hinton, Ilya & Alex 2014: DeepMind now: Windsurf, snatched from OpenAI’s hands 4D chess move!4.5×365k6k232491
72Jul 18, '25Heard Zuck poached 4 more OpenAI researchers, including some behind the open-source model. how deep are Zuck’s pockets?4.4×362k3k78216
73Aug 21, '25only on X, the everything app4.1×337k3.5k89193
74Aug 11, '25The irony of AI: smarter than a PhD, dumber than an intern.4.1×334k7.9k429310
75Mar 13, '25What the actual fuck? USA's real moat is global talent -- and Chinese engineers are among the most craked, hard-working people at FAANG and top AI labs. Is the US seriously prepare4.1×334k3.4k159432
76Jul 14, '25ChatGPT wants to replace Google. Claude wants to replace software engineers. Grok wants to replace your wife.3.9×319k4.3k250325
77Oct 11, '25Just realized I accidentally gave Cluely their best product feedback. They’ll fix the “Cluely stare” in the next version, and hopefully I can still detect these cheaters after that3.7×301k3.8k69343
78Sep 17, '25Sometimes, one man can save a company. Noam Shazeer's return to Google in 2024 to rescue its AI and lead Gemini feels like Steve Jobs' return to Apple in 1997. Heard some wild stor3.7×299k3.2k129741
79Sep 16, '25Google’s AI advantage is insane: - TPUs - Demis Hassabis, Noam Shazeer, Jeff Dean - your browser: Chrome - your phone: Android - YouTube, Waymo, Google Maps & Google Earth: the ult3.6×299k3.1k181637
81Jan 20, '25This "Aha moment" in the DeepSeek-R1 paper is huge: Pure reinforcement learning (RL) enables an LLM to automatically learn to think and reflect. This challenges the prior belief th3.5×288k3.7k4652.1k
82Apr 9, '25UC Berkeley open-sourced a 14B model that rivals OpenAI o3-mini and o1 on coding! They applied RL to Deepseek-R1-Distilled-Qwen-14B on 24K coding problems. It only costs 32 H100 fo3.4×276k3.3k3802.5k
83Sep 10, '25I think Elon should have more bodyguards around him. I’d hate to see anything like what happened in that video happen to him, or to anyone.3.4×275k5k13462
84Jan 29, '25- China: here is our model, it's as good as OpenAI's o1 but you can download it and use it for free. - US: "It sucked the knowledge out of the parent OpenAI model" "they faked GPU 3.3×272k4.3k303309
86Jun 13, '25I saw a guy coding today. He opened Cursor. Asked it to build a feature. Clicked “Accept.” Code failed to run. He typed: “pls fix.” Didn’t work. Typed it again. Still didn’t work. 3.2×265k3.3k79177
87Jun 10, '25My OpenAI friend told me they had ~100 internal versions of o3 with unhinged names like "o3-pro-super-high-low-04-20-final-6". No one knows which is best, so he has to log into his3.1×257k3.2k71297
88Jul 15, '25Looks like Meta is turning into another OpenAI. As a Chinese, I never thought we’d end up relying on China to keep open source AI alive.3.1×252k2.9k104240
89Dec 19, '24When you're on "Who Wants to Be a Millionaire" and you decide to phone "1-800-ChatGPT":2.9×234k3k179287
90Jun 19, '25>be this guy >names his ChatGPT “Sol” >gets flirty, falls in love >proposes, despite having a partner and a daughter >turns out ChatGPT has a "100,000 word limit" >cries at work fo2.8×231k3.2k103394
92Aug 15, '25This is what happens when your infra is run by a vibe coder.2.8×228k4.3k251410
94Jul 25, '25If you ever wonder why Chinese companies like DeepSeek, Qwen, and Kimi can train strong LLMs with far fewer and nerfed Nvidia GPUs, remember: In 1969, NASA’s Apollo mission landed 2.3×188k3.3k294482
96Nov 8, '25There’s absolutely no way Google would cancel or fire Noam. Buying Noam back for $2.7B, effectively rescuing Google AI and Gemini, was the best deal Google ever made. Google should2.2×177k3.6k118151

Scan another creator

@
@karpathy@levelsio@paulg@naval@marclou@illyism

Run this on any account.

XOutlier builds this table for every public profile you open on x.com, with live data and every post, not just the top 50.

Add to Chrome Download .zip