❄️Andrew Zhao❄️ @_AndrewZhao
PhD @Tsinghua_Uni. Absolute Zero,ExpeL,Diver-CT Ex. intern@MSFTResearch,@ BIGAI. Interested in RL, Reasoning/Safety 4 LLMs, Agents. On industry job market 2026 andrewzh112.github.io Joined September 2020-
Tweets1K
-
Followers4K
-
Following3K
-
Likes3K
Today, @ekindogus and I are excited to introduce @periodiclabs. Our goal is to create an AI scientist. Science works by conjecturing how the world might be, running experiments, and learning from the results. Intelligence is necessary, but not sufficient. New knowledge is…
iykyk arxiv.org/pdf/2509.24527
In case you didn't know my recent work Single-stream Policy Optimizaton (SPO), a group-free low variance policy gradient algorithm. Check this blog out: zhongwenxu.notion.site/Single-stream-… and paper: arxiv.org/abs/2509.13232
In case you didn't know my recent work Single-stream Policy Optimizaton (SPO), a group-free low variance policy gradient algorithm. Check this blog out: zhongwenxu.notion.site/Single-stream-… and paper: arxiv.org/abs/2509.13232
🌀New work: Era of Real-World Human Interaction 🌀 📝: arxiv.org/abs/2509.25137 - RL *directly* from User Conversations - Organic replies + long-term history are learning signal - Trained on WildChat, beats RLHF at *user* level -> the future for personal Super Intelligence? 🧵1/6
After the crazy 极GRPO weekend, let's get rid of the scalar reward or any policy optimization related to it. We explored learning from *verbal feedback* and obtained interesting results:
After the crazy 极GRPO weekend, let's get rid of the scalar reward or any policy optimization related to it. We explored learning from *verbal feedback* and obtained interesting results:
🚀 Introducing DeepSeek-V3.2-Exp — our latest experimental model! ✨ Built on V3.1-Terminus, it debuts DeepSeek Sparse Attention(DSA) for faster, more efficient training & inference on long context. 👉 Now live on App, Web, and API. 💰 API prices cut by 50%+! 1/n
I don’t often tweet on technical topics but I may have an opposite opinion here…
NO verifiers. NO Tools. Qwen3-4B-Instruct can match DeepSeek-R1 and o3-mini (high) with ONLY test-time scaling. Presenting Recursive Self-Aggregation (RSA) — the strongest test-time scaling method I know of! Then we use aggregation-aware RL to push further!! 📈📈 🧵below!
(1/x) Ever had your #LLM-#RL training mysteriously collapse? 📉 You're not alone. We saw #agentic RL runs fail with exploding #gradients, and found the culprit: a fundamental "training-inference mismatch." Our new #blog post demystifies this vicious cycle.…
😠💢😵💫Tired of endless data collection & fine-tuning every time you try out VLA? Meet RDT2, the first foundation model that zero-shot deploys on any robot arms with unseen scenes, objects & instructions. No collection. No tuning. Just plug and play🚀 Witness a clear sign of…
One thing people rarely mention about research: ideas set the upper bound of your work, but debugging sets the lower bound. Universities teach us how to chase impactful ideas, but they rarely teach us how to debug large, messy ML systems. Here are a few principles I found useful…
This is the command most people should run when using codex-cli codex --search --model=gpt-5-codex -c model_reasoning_effort="high" --sandbox workspace-write -c sandbox_workspace_write.network_access=true
The field of robotics is undergoing a historic revolution right now. I’ve spent the last year thinking about how to mentally model the breakneck progress in robotics + AI. With the help of mascots like “The AGI Bro”, we can try to sift through the noise 🧵
Proud to have been part of the team behind Gaia2 and ARE! ARE = a gym/platform for scaling up LLM agent envs for evals & RL Gaia2 = a new benchmark for hard & practical agent tasks (search, execution, ambiguity, time, noise, & multi-agent) tinyurl.com/aregaia2
Most agent benchmarks assume static, perfect worlds. But real life is asynchronous, noisy, and ambiguous. 🌍 🚀 Meet Gaia2 + ARE: a new benchmark and open-source platform for creating environments and evaluating AI agents in (more) realistic environments.

Chuanyang Jin @chuanyang_jin
470 Followers 390 Following PhD @JohnsHopkins | Intern @AIatMeta FAIR ⏰ Past: @MITCoCoSci & @MIT_CSAIL & @nyuniversity
Sushil Pokhrel @sushilpokhrel
3K Followers 6K Following Biomedical Engineering researcher turned Systems Designer, Machine learning, ai +Robotics+systems +design ,cryptography etc. Felt in love with ML
Andrew Liao @AndrewOail
5 Followers 159 Following
Ruochen Zhang @ruochenz_
797 Followers 2K Following Interning @cohere, PhDing @Brown_NLP & @health_nlp, working on multilingual NLP and interpretability. Prev: Undergrad @sutdsg, she/they
Eduardo C. Garrido-Me... @vedugarmer
465 Followers 694 Following Doctor Ingeniero en Informática. Profesor investigador en ICADE-IIT @UCOMILLAS. Trabajo en Inteligencia Artificial. Me gusta pasear con mis hijos y la lectura.
yonromai @yonromai
4 Followers 236 Following
hyeju defender @olhye_supremacy
24 Followers 2K Following
Bhaskar Jha @hmmbhaskar
6 Followers 252 Following
Marc Pinet @marcpinet0
5 Followers 144 Following PhD Research Student in AI @Orange - @Orange_Future | Self-Supervised Deep Learning for #AnomalyDetection & #Explainability in #TimeSeries
Benjamin Ruppik @ben300694
144 Followers 3K Following Topological Deep Learning for Natural Language Processing and Dialogue Systems @ Heinrich-Heine-University Düsseldorf @HHU_de
Nilotpal Mishra @NilotpalMi85217
4 Followers 54 Following
Sepehr Heidari @SepehrHeidari81
340 Followers 7K Following Interested in Physics🔭, Mathematics🧮, Computer Science 💻, and Dad Jokes🥸 https://t.co/jtKArilTEr
Maxime Guerreiro @punkeel
2K Followers 2K Following principal software engineer @ cloudflare. Clouds are my owns
Anxiety @LiElbert76127
19 Followers 322 Following
yyspirky @yyspirky
71 Followers 777 Following
Terry Chen @tchenml
129 Followers 1K Following
Natasha Jaques @natashajaques
31K Followers 1K Following Assistant Professor @uwcse and Staff Research Scientist at @GoogleAI. Let's get off this app: https://t.co/jbH2oAjbPN
Benhao Huang @huskydogewoof
92 Followers 744 Following M.S. student @mldcmu, Prev. @sjtu1896 | Opinions approved by my puppy.
mouse mikey @mousemikey36341
0 Followers 74 Following
Matt Sheehan @mattsheehan88
17K Followers 6K Following US/China AI & tech. Fellow @CarnegieEndow. Author "The Transpacific Experiment: How China & California Collaborate & Compete For Our Future." Nuggets.
SanchosDonkey @sanchos_donkey
3 Followers 334 Following
orko @_orkorko
67 Followers 578 Following
U @dogeornothin
223 Followers 462 Following
Michael Hanegan @mhanegan
1K Followers 5K Following Founder, Center for the Future of Learning and Work || Professor of AI and Work || Co-Author of “Generative AI and Libraries” (published by ALA) || Generalist
Philippe Laban @PhilippeLaban
1K Followers 702 Following Research Scientist @MSFTResearch. NLP/HCI Research.
Ekue Kpodar @ekpodar
4K Followers 4K Following Building at the edge of AI, marketing, and complex systems. 🚀 Sharing experiments.
Jay @jayvaaty
0 Followers 418 Following
Ellevv @manaanaf331
349 Followers 275 Following
Pretan gorges @GorgesPr20
333 Followers 6K Following
Changlong Yu @noever0812
18 Followers 519 Following
Rod Mamin (🌍🚀�... @0xIonRod
10K Followers 6K Following 🚀 Degen Space Engineer | 🌕 Founder @LunCoSim | 🧮 Mathematician | 🪄 Hyperstructures ℋ Summoner @hyperdesci | 🌍 DeSci | DeSpace
Yteguiv @Yteguiv553781
127 Followers 3K Following
Dir Ha Tan @dirhatan55
627 Followers 6K Following Unlocking the mysteries of Celtic cultures! Ethnologist & linguist unraveling languages & traditions.
Aida @Aida1085111Aida
1 Followers 57 Following
Khalali Johnson @Tru3senseof
176 Followers 864 Following 🕊️Christ ❤️ Crypto • BTC • NVDA • PLTR • TSLA
AnalyticsData @AnalyticsDataCo
397 Followers 2K Following
Herbert R. Sim @HerbertRSim
403K Followers 338K Following CMO @AICEAN_AI 📹 VC https://t.co/zjtLNJnVNp 📊https://t.co/ezk6p49ePk (1997) 🤖 https://t.co/oNUDO3nhNJ (1999) 🧠 https://t.co/QgASuahnNZ (2002) 🧬 @XcomErc20 ⚔️ | https://t.co/FJGx2ilPTd
X Freeze @amXFreeze
36K Followers 1K Following Tech updates, strategy, and bold takes. I am the coolest villain, don't forget
Killian Sheriff @KillianSheriff
513 Followers 3K Following @PeriodicLabs, Ph.D. @MIT | Prev @ToyotaResearch @mcgillu 21' | From 🇫🇷 | Doing Materials Science + AI research to better understand high-entropy alloys.
Reiichiro Nakano @reiinakano
7K Followers 977 Following i like building awesome things with awesome people 🇵🇭 🇯🇵
Ekin Dogus Cubuk @ekindogus
5K Followers 432 Following Co-Founder of @periodiclabs Past: Lead of materials science and chemistry at @GoogleDeepMind; Google Brain
Prafulla Dhariwal @prafdhar
44K Followers 520 Following Technical fellow @OpenAI. Co-creator of GPT-4o, GPT-3, DALL-E 2, Jukebox, Glow, PPO. Previously @MIT '17
Jihoon Tack @jihoontack
728 Followers 678 Following Incoming Senior Researcher at @MSFTResearch | PhD @kaist_ai | Google PhD Fellow | Prev: Visiting @oxcsml, Research Intern at FAIR @AIatMeta
Mike Krieger @mikeyk
463K Followers 267 Following Chief Product Officer at @anthropicai. Before: co-founder & CTO of @instagram and @artifact_news
Rishi Mehta @rishicomplex
3K Followers 286 Following Solve i̶n̶t̶e̶l̶l̶i̶g̶e̶n̶c̶e̶ ̶ coding, use it to solve everything else | Research @AnthropicAI | Past: RL @GoogleDeepmind: AlphaProof co-lead, Gemini.
Xiaokang Chen @PKUCXK
2K Followers 28 Following Researcher @deepseek_ai | Previously Ph.D at Peking University @PKU1898 Projects: #JanusPro, #DeepSeekVL2
Bill Chen @realchillben
2K Followers 580 Following @openai ; Prev @ycombinator @Retool @Meta ML @Columbia
Siddarth Venkatraman @siddarthv66
609 Followers 474 Following PhD at Mila | RL and other stuff I find interesting
Hexiang Hu @hexianghu
2K Followers 694 Following Multimodal @xAI: Cooking models for grok chat & imagine Prev: gemini 1 / 2 & imagen 3 @GoogleDeepMind.
Songming Liu @songming_liu
605 Followers 88 Following CS PhD at @Tsinghua_Uni, focusing on building large-scale robotic datasets and training large models for generalizable robotic manipulation.
Andrey Kolobov ✈️... @Andrey__Kolobov
456 Followers 73 Following lead, robot learning @ Microsoft Research // researcher of AI decision-making for embodied agents // aviation fan // all opinions are my own
Samir @_samirism
6K Followers 910 Following ChatGPT Personalization and Memory @OpenAI , previously Eng @Snap
Christina Wadsworth K... @ChristinaHartW
8K Followers 175 Following leading personalization @OpenAI | previously @Meta, @Instagram | SF
Snowflake @Snowflake
59K Followers 1K Following Snowflake delivers the #AIDataCloud to help leading organizations share data, build applications and power their business with AI.
X Freeze @amXFreeze
36K Followers 1K Following Tech updates, strategy, and bold takes. I am the coolest villain, don't forget
Yiwen Yuan @yiwenyuan98
6K Followers 98 Following @xai | prev @kumo_ai_team @palantirtech @CarnegieMellon
Zijian Hu @zijianhu
64 Followers 296 Following ML researcher at @scale_AI. Ex-@tiktok_us. @USC/@CSatUSC Alumni. Working on #LLM
Alaa El-Nouby @alaa_nouby
844 Followers 390 Following Research Scientist at @Meta . Previous: @Apple, @Inria, @MSFTResearch, @VectorInst and @UofG
Eliezer Yudkowsky @allTheYud
3K Followers 17 Following High-volume account of @ESYudkowsky, the original AI alignment guy. If it's missing punctuation, it's humor. If you can't tell, it's probably also humor.
Da Yu @DaYu85201802
507 Followers 148 Following Research Scientist at Google Research. Former intern at @MSFTResearch and @GoogleAI. Joint PhD between Sun Yat-sen University and Microsoft Research Asia.
slime @slime_framework
215 Followers 3 Following The LLM post-training framework for RL Scaling. https://t.co/4ILpx8hfKN
Lily Lim @LiLiDuc22
3K Followers 220 Following xAI Head Legal Eagle: Lily is an adventurer, former rocket scientist, and now launcher of products at the innovative Elon Musk AI start-up, xAI.
Andree Jacobson @nmswede
11K Followers 1K Following Building massive scale HPC and AI compute at @xAI. Strong @Grok supporter. Work and random stuff. Views are my own.
Arno @aarnogau
9K Followers 198 Following grok web @xAI | prev founder @iudexai | prev eng @scale_AI, @googlex
Lianmin Zheng @lm_zheng
14K Followers 620 Following Member of technical staff @xAI | Prev: Ph.D. @UCBerkeley, Co-founder @lmsysorg
Ying Sheng @ying11231
12K Followers 732 Following @lmsysorg @sgl_project | Prev. @xAI @Stanford | Assist Prof @UCLA. (delayed) | Do it anyway | Live to fight another day
Ali Panju @alipanju_
3K Followers 149 Following special projects @perplexity_ai, prev @socialcapital
Ulyana Piterbarg @ulyanapiterbarg
944 Followers 630 Following reasoning, agents, RL, + open-endedness | PhDing at @nyuniversity and @AIatMeta, prev @MIT
Romain Froger @froger_romain
128 Followers 237 Following PhD @AIatMeta, MSL Agents and @Inria. @GeorgiaTech & UTC alumni.
President Donald J. T... @POTUS
3.6M Followers 4 Following 45th & 47th President of the United States. The Golden Age of America Begins Right Now.
Bing Liu @vbingliu
842 Followers 98 Following Director of Research @Scale_AI. Prev: GenAI @Meta, PhD @CarnegieMellon.
Hattie Zhou @oh_that_hat
10K Followers 853 Following I want to understand things deeply and explain them well. Building friendly AI @AnthropicAI Give me anonymous feedback: https://t.co/7aBNrpbad8
Yongchao Zhou @Yongchao_Zhou_
3K Followers 420 Following Build Intelligence @xai | ML PhD @UofT @VectorInst | Prev. @GoogleAI @GoogleDeepMind
Heng-Tze Cheng @HengTze
2K Followers 175 Following Research Director & Principal Scientist @GoogleDeepMind, Gemini Team | Lead of LaMDA LLM & AI | Worked on Duplex, TensorFlow, Wide & Deep Learning | Hiring!
M @m
7K Followers 0 Following
Vincent Stark @theVincentStark
2K Followers 109 Following Making Grok safe for users and beneficial for society