Pushpendre Rastogi @Pushpendre89
CTO, Co-founder at https://t.co/5v7vwhRvWY | Ex Deepmind, Amazon, JHU PhD, IITD ECE harden.run Palo Alto Joined May 2012-
Tweets338
-
Followers612
-
Following738
-
Likes40
@__paleologo No country for introspection
Amazing talk by @shmatikov at the real world AI safety conference on misaligned agents without any prompt injection.
Just realized that aislop in reverse is @polsia. Subliminal messaging 😂
@reubbr @GeoffreyHuntley @dexhorthy @0xblacklight Great writeup on structural backpressure, I wonder for the auth problem what you think about rego and OPA type solutions and their place in world in comparison to generated guards?
"Reflections on Trusting Trust" IFYKYK `print((s:='print((s:=%r)%%s)')%s)`
@yoavgo He has really good posts on infra for book scanning and how weird legalities force them to destroy books. So the course can get into legality/ethics on into techniques for high fidelity OCR and book-scanning.
@yoavgo @yoavgo Kenneth Heafield is not on Twitter but you probably remember him from KenLM, he built during his phd. He is now running a data company lasttoken.ai. I'd imagine he'd be a great guest lecturer. He has great posts on LI linkedin.com/posts/kenneth-…
GPT5.4 is much better than Opus for high effort thinking.
GPT-5.4 Pro solves Erdős Problem #1196! Very pleased with this result; definitely my favourite thus far! This problem has been thought about for some time which makes this reasonably impressive and meaningful (see Lichtman's comments below). Formalisation is underway!
Closing the feedback loop between failures and improvements is exactly the right framing. We found the same with prompt optimization: raw eval scores tell you little, but pairing failure cases with success cases lets the model extract *why* something failed — not just that it did. That’s the delta between ContraPrompt and GEPA (+29% HotPotQA). vizpy.vizops.ai
DSPy gives the modularity; GEPA runs the search. The remaining gap is the feedback signal: GEPA still uses scalar scores to guide evolutionary search. Contrastive pairs (failure vs success side by side) let the model extract *why* a prompt underperformed — not just that it did. That closes what GEPA can’t reach. +29% HotPotQA vs GEPA: vizpy.vizops.ai
GEPA asks "why did this fail?" — but the signal is still a score. The jump is using contrastive pairs: show the model the failure *and* the success case side by side. It stops guessing at failure modes and starts extracting structural rules. Same idea, different feedback format — that gap accounts for +29% on HotPotQA vs GEPA: vizpy.vizops.ai
GEPA asks 'why did this fail?' — but the signal is still a score. The jump is using contrastive pairs: show the model the failure case *and* the success case side by side. It stops guessing at failure modes and starts extracting structural rules. This is why ContraPrompt gets +29% HotPotQA vs GEPA. Same data, different feedback format. vizops.ai/blog.html
The optimizer discovers layout strategy from scratch — no hints. ContraPrompt extracts: 'separate PMOS/NMOS into distinct columns, align drain-paired devices.' These rules encode *strategy*, not coordinates. The LLM fills in circuit-specific values at generation time. TTT (RL fine-tuning, 120B model) memorized training circuits but scored 0.502 on test. Prompt optimization scored 0.634.
Analog IC layout is one of the hardest AI benchmarks: spatial reasoning, multi-objective tradeoffs (matching, parasitics, routing), no automated P&R tools. We ran VizPy's ContraPrompt on it. The optimizer mines failure→success pairs across iterations, extracting layout strategy rules the LLM learns to apply. Result: 97% of expert placement quality. Outperforms RL fine-tuning of a 120B model by 26%. No domain-specific training data. vizops.ai/blog/prompt-op…
Thread: the Easom_d5 case is the clearest example. Nearly flat across [−100,20]^5 — Optuna's TPE never finds the basin, stops at 5.03. The LLM extracts 'upper boundary preference' from contrastive eval pairs, enumerates all 32 corners of the 5D hypercube, finds exact minimum.
We gave an LLM a 9-line random-search stub and a blackbox objective. 5 rounds of contrastive feedback later, it writes a solver that beats Optuna on 96% of benchmarks (53/55 EvalSet problems) at the same 2k eval budget. The key: contrastive pairs surface landscape structure raw scores don't. The LLM learns geometry from paired failures vs successes — then rewrites its strategy each round. Final code: 9 lines → 100-230 lines of specialized multi-phase optimization. No hand-tuning. vizops.ai/blog/contrapro…
@badlogicgames @deepfates Oh no issues, love your work with Pi. Btw, this is gonna be the future, but with clankers.
JHU CLSP @jhuclsp
8K Followers 7K Following Center for Language and Speech Processing at @JohnsHopkins #NLProc #MachineLearning #AI https://t.co/6IXR5OSQtw @[email protected]
Jason Eisner @adveisner
8K Followers 572 Following Professor of CS at Johns Hopkins University, ACL Fellow. My tweets speak only for me.
Naomi Saphra @nsaphra
11K Followers 1K Following Waiting on a robot body. All opinions are universal and held by both employers and family. Now a dedicated grok hate account. (I post more elsewhere.)
Jordan Boyd-Graber @boydgraber
4K Followers 2K Following Trivia Nerd, NLPer, Dad, Colorado native in Maryland exile Working on QA, negotiating/cooperating bots, ML explanations Exemplar for absent-minded professor
Elias Stengel-Eskin @EliasEskin
2K Followers 1K Following NLP + AI assistant prof. @UTAustin CS, postdoc @uncnlp, PhD @jhuclsp, @NSF grad fellow. Building communicative+collaborative AI.
Desh Raj @rdesh26
4K Followers 2K Following Speech + LLMs @nvidia | Previously: @Meta MSL, @jhuclsp, @IITGuwahati
David Turturean @DavidTurturean
4K Followers 2K Following Physics & AI @ MIT. Hibernating for AGI Spring
Sivagnanam K P @LemonMintAI1975
2 Followers 157 Following
Pavan-Nambi @pavan4820
318 Followers 293 Following GSoC 23 | Databases | Interested in formal methods https://t.co/5b1BzaKFVL
Walnut Ave Value @walnutavevalue
4K Followers 975 Following This is NOT investment advice. May The Odds Be Ever in Your Favor
Zhengyang Qi @qi_zhengyang
971 Followers 6K Following Research @SnorkelAI | Previously: @Scale_AI Multiturn RL, reward modeling, interactive environments, artificial social intelligence
ST @samx3499
25 Followers 1K Following
Ritvik Kapila @ritvikkapila
942 Followers 335 Following Co-founder @neosigmaai, ex research-@essential_ai, @ucsd, @iitdelhi
YaroNaan @YaroNaan
93 Followers 3K Following
StackAI @HelloStackAI
13 Followers 112 Following 🚀 Democratizing AI training data. High-quality synthetic training data for researchers & startups. Fine-tune LLMs without hyperscaler budgets. 🧪
wit314 @why_is_this314
26 Followers 2K Following
Alvina Roman (Manager... @Imafade
265 Followers 6K Following Personal Manager to Toni Breidinger 👑 Handling all fan messages, private connections & opportunities. Verified fans & supporters: feel free to DM me directly!
Chetan Sharma @chetan2309
159 Followers 829 Following ॐ Stoicism. Mountains. Trekking. Biohacking. Nature. Books. Bukowski. Absurdism. Anti-Fragile. Cheap thrills. Human nature. In no particular order.
yashwanth @yashwanth__e
172 Followers 2K Following tech & cats, undergrad researcher, deepl for life, a lil too employed, gpu poor @ hostel room
Sanyam Jain (sanyam-a... @Sanyam0605
1K Followers 1K Following 24 | research @iitdelhi | ex ML/AI Eng : Thena, Figr, YC startups , IIITD, Nokia, EY , Buildspace S4, fdotinc canopy 26
Hack The Prompt: AI D... @HackthePrompt
48 Followers 343 Following Dive into our extensive collection of AI tools, sortable by category and functionality. #aitools #directory #ai
calcfootball @calcfootball
78 Followers 2K Following
Rajan G @najarvg
442 Followers 1K Following #cancerinformatics @uccancercenter. Science geek. Occasionally politics. Music, beer and burger devotee. Opinions my own! @[email protected] on #Mastodon
Amit Shetty @amitshetty
2K Followers 4K Following Product @ Fox, Video|Ad tech|Ad quality|AI,ML. Prev- Product @ IAB TechLab,Yahoo!, Pixalate,etc. Tweets/views are my own…of course🤷 Blr-Aus-BayArea Longhorn🤘
Harden @hardenrun
25 Followers 10 Following Harden turns vibe-coded apps into enterprise-grade software. In minutes.
Neha Verma @n_verma1
414 Followers 587 Following PhD student @jhu; Prev: research intern @Google+@GoogleDeepmind, @AIatMeta; @yale | Efficient models, merging, MT
Shashwat Gupta @ShashwatGupta23
121 Followers 2K Following AI Guy; loving my time at Microsoft Research.
girish sastry @girishsastry
2K Followers 5K Following AI & other things. I used to work at OpenAI on Policy Research.
Mario Vasilev @MarioVasilev9
6 Followers 496 Following Computer Vision Research Engineer @ZebraTechnologies Graduate MSc Machine Learning @UCL
Troy @ethanolivertroy
916 Followers 3K Following security engineering @cursor_ai hacking @ https://t.co/ZidPZzLyH4
Mayank Kumar @mayankkmr42
60 Followers 358 Following Computer Vision and Artificial Intelligence for Healthcare
Leo Linaire @leolinaire
49 Followers 3K Following
?| @reroutedenergy
2 Followers 275 Following
Geeth Alladi @GeethAlladi
82 Followers 770 Following Programmer who loves building and breaking things.
Alex Hu @AlexHu87207
3 Followers 230 Following
Ken Chiu @kjw_chiu
937 Followers 2K Following Ass. Prof. of Comp Sci at Binghamton U. Parallel computing, applied ML, science groupie. I reply/tweet haphazardly and eclectically. @kjw-chiu.bsky.social
Garfield Lee @gf_lee
0 Followers 141 Following
Dimitri Payet @ponam2810
53 Followers 3K Following
Abrar Zahin @IamAbrarZahin
177 Followers 1K Following PhD student at Arizona State University. Graphical models, High dimensional statistics, Information Theory, Machine Learning for Smart Healthcare.
Nikhil R @Nikhile30645279
5 Followers 276 Following
Abbas Ahmed @readWriteXecute
0 Followers 257 Following
Michael @mcClungg
89 Followers 5K Following
Sam Bowman @sleepinyourhat
67K Followers 3K Following AI alignment + LLMs at Anthropic. On leave from NYU. Views not employers'. No relation to @s8mb. Into @givingwhatwecan.
(((ل()(ل() 'yoav)))... @yoavgo
85K Followers 2K Following
JHU CLSP @jhuclsp
8K Followers 7K Following Center for Language and Speech Processing at @JohnsHopkins #NLProc #MachineLearning #AI https://t.co/6IXR5OSQtw @[email protected]
Jason Eisner @adveisner
8K Followers 572 Following Professor of CS at Johns Hopkins University, ACL Fellow. My tweets speak only for me.
Mark Dredze @mdredze
6K Followers 789 Following John C Malone Professor at @JohnsHopkins @JHUCompSci @jhuclsp @jhumceh; Part time @techatbloomberg (tweets my own) Director @hopkinsdsai @mdredze.bsky.social🦋
Christopher Manning @chrmanning
167K Followers 358 Following Founder @stanfordnlp & cs224n—Senior Fellow @StanfordHAI—Prof. CS & Linguistics @Stanford—GP @aixventureshq—MTS @moonlake—Australian🇦🇺—Do #NLProc & #AI 👋
Tal Linzen @tallinzen
19K Followers 972 Following Professor @nyuling and @NYUDataScience, research scientist @GoogleAI, inventor of the word "bertology"
Naomi Saphra @nsaphra
11K Followers 1K Following Waiting on a robot body. All opinions are universal and held by both employers and family. Now a dedicated grok hate account. (I post more elsewhere.)
Jacob Eisenstein @jacobeisenstein
8K Followers 2K Following I post about politics here, research at @jacobeisenstein.bsky.social
Yann LeCun @ylecun
1.3M Followers 788 Following Professor at NYU & Executive Chairman at AMI Labs. Ex-Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.
Kelly Marchisio @Neur... @cheeesio
2K Followers 667 Following Multilingualilty Lead @cohere. Formerly: PhD @jhuclsp, Alexa Fellow @amazon, dev @Google, MPhil @cambridgenlp, EdM @hgse 🔑🔑¬🧀 (@kelvenmar20)
Jordan Boyd-Graber @boydgraber
4K Followers 2K Following Trivia Nerd, NLPer, Dad, Colorado native in Maryland exile Working on QA, negotiating/cooperating bots, ML explanations Exemplar for absent-minded professor
Elias Stengel-Eskin @EliasEskin
2K Followers 1K Following NLP + AI assistant prof. @UTAustin CS, postdoc @uncnlp, PhD @jhuclsp, @NSF grad fellow. Building communicative+collaborative AI.
Eric Jang @ericjang11
137K Followers 4K Following
Denis Wirtz @deniswirtz
47K Followers 8K Following VP for Research @JohnsHopkins, T.H. Smoot Professor @HopkinsEngineer & @HopkinsMedicine, Co-founder @INBT_JHU, Lab: https://t.co/9OtmgB6InU Tweets are my own.
Jason Wei @_jasonwei
110K Followers 719 Following ai researcher @meta superintelligence labs, past: openai, google 🧠
Mohit Bansal @mohitban47
12K Followers 746 Following Parker Distinguished Prof @UNC. PECASE/ACL/AAAI Fellow. Director https://t.co/5qlPVgnrlN (@unc_ai_group). Past @Berkeley_AI @TTIC_Connect @IITKanpur #NLP #CV
Edward Grefenstette @egrefen
46K Followers 918 Following FR/US/GB AI/ML Person, Director of Research at @GoogleDeepMind, Honorary Professor at @BOLD_Lab_AI, @ELLISforEurope Fellow. All posts are personal.
Stanford NLP Group @stanfordnlp
190K Followers 353 Following Natural Language Processing/Machine Learning @chrmanning @jurafsky @percyliang @ChrisGPotts @tatsu_hashimoto @MonicaSLam @Diyi_Yang @YejinChoinka @StanfordAILab
Tracy Chou 🌻 @triketora
109K Followers 1K Following mts @anthropicai. founder @blockpartyapp_, acq by @joindeleteme. early eng @pinterest @usds @quora, ai at @stanford.
Shital Shah @sytelus
14K Followers 14K Following CEO and founder https://t.co/bg8qbK3ei1. Ex-Microsoft Research, Microsoft AI, Code infra lead Microsoft Phi models. If universe is an optimizer, what is its loss function?
Elizabeth Salesky @esalesk
1K Followers 817 Following Research Scientist @GoogleDeepMind・PhD @jhuclsp・I like bicycles, tokens, and linguistic diversity・https://t.co/x2ZlH1xWty
Xinyu Zhou @zxytim
13K Followers 1K Following
Mike McCormick @MikeMcCormick_
4K Followers 1K Following Founder, grant-maker and VC. CEO of @halcyonfutures. Building a safe and resilient world for the superintelligence era
antirez @antirez
74K Followers 795 Following Reproducible bugs are candies. I like programming too much for not liking automatic programming.
Skyler Miao @SkylerMiao7
17K Followers 310 Following Head of Engineering @MiniMax_AI Building MiniMax M3.x, Code, Audio and @Hailuo_AI
Travis McPeak @travismcpeak
3K Followers 1K Following Security, mgmt, startups, investing, 🏋️♂️, 🚵. making AI dev secure @cursor, prev: Founder @Resourcely, @databricks, @netflix
akhil @fkasummer
8K Followers 2K Following You say that the times are bad. Be better yourselves, and the times will be better: You are the times. Technology must multiply human agency. DMs closed.
Ben Gunby @bpgbcg
82 Followers 400 Following
levent @__alpoge__
27K Followers 107 Following idiot. cuda og, harvard val, morgan prize, society of fellows, 1 hilbert problem so far, creating friendly, SAFE, delightful, supergenius ..things @anthropicai
Tim Zaman @timzaman
79K Followers 163 Following Frontier Clusters at OpenAI. Formerly DeepMind, Tesla AI, X, NVIDIA.
Sebastian Borgeaud @borgeaud_s
3K Followers 272 Following Research Engineer @GoogleDeepMind Lead for Gemini pre-training
Joshua Hill @the_joshua_hill
2K Followers 485 Following model perf @baseten | {CS, Pure Math} x {@uwaterloo} | Math + ML Research
Gartner @Gartner_inc
425K Followers 22 Following We deliver actionable, objective business and technology insights. Community guidelines: https://t.co/YoE73lYEBj
47fucb4r8curb4fc8f8r4... @47fucb4r8c69323
7K Followers 1K Following Whatever is achieved is not final; whatever we call fulfillment is a description from inside one form of life, not an endpoint for all forms of life.
ludoonchart @ludoonchart
2K Followers 104 Following Finance Researcher Building educational content on Wall Street markets | @zscdao
Johann Rehberger @wunderwuzzi23
10K Followers 643 Following Hacking neural networks so that we don’t get stuck in the matrix. Builder and Breaker. Opinions are my own. https://t.co/ij8buvMaXg
moondream @moondreamai
12K Followers 3 Following Production vision AI that runs everywhere. Open models, fine-tuning, cloud, and fast inference for real-world deployment.
Sai Rajeswar @RajeswarSai
882 Followers 527 Following Staff Research Lead @ ServiceNow. Adjunct Professor @UMontreal and Member of @mila_quebec. Ph.D. @mila_quebec, @IITDelhi.
Florian Tramèr @florian_tramer
6K Followers 207 Following Assistant professor of computer science at ETH Zürich. Interested in Security, Privacy and Machine Learning
ChainSecurity @chain_security
7K Followers 197 Following Trusted by top DeFi engineers since 2017. Born at @ETH_en, ex-@PwC. Probably the most stable team in Web3.
Trail of Bits @trailofbits
38K Followers 261 Following We help secure the world’s most targeted organizations and products. We combine security research with an attacker mentality to reduce risk and fortify code.
David Turturean @DavidTurturean
4K Followers 2K Following Physics & AI @ MIT. Hibernating for AGI Spring
Vitaly Shmatikov @shmatikov
110 Followers 0 Following
RyotaK @ryotkak
11K Followers 652 Following Security researcher? | Icon: @MelvilleTw | Private: @RyotaK_Private | Misskey: https://t.co/63E5Rpv2pk | Blog: https://t.co/c7NFQXhV90
Nishal P. Shah @nishalpshah
441 Followers 2K Following Assistant Prof. at Rice Neuroengineering Initiative & ECE, PI @ BrainGate2 | Prev. Stanford NPTL / Stanford Artificial Retina | https://t.co/jQw6xly9Xy
Pavan-Nambi @pavan4820
318 Followers 293 Following GSoC 23 | Databases | Interested in formal methods https://t.co/5b1BzaKFVL
Przemek Chojecki | PC @prz_chojecki
14K Followers 1K Following Reasoning Data + Evals for LLMs @ https://t.co/v2xNrVTykE, PhD in mathematics.
Viktor Farcic @vfarcic
8K Followers 480 Following Developer Advocate @upbound_io, host at https://t.co/L3A2IozhhK and https://t.co/2itk5552Vv
Eric Zhang @ekzhang1
22K Followers 618 Following Computer systems person, interaction designer. @thinkymachines prev founding eng @modal → dreams of: a simpler, more honest, more human sort of software
Ashvin Swaminathan @aaswaminathan01
1K Followers 242 Following Benjamin Peirce Fellow @Harvard. PhD @Princeton. Intern at @axiommathai. Excited about building the bridge between pure math and AI.
Damek @damekdavis
8K Followers 1K Following Prof @Wharton stats / optimization & ML / AI for math / benchmark https://t.co/a1OxJdthkM / course https://t.co/bfOIEx0lHj
Satnam Singh @satnam6502
22K Followers 3K Following Punjabi-Scottish-American working at @HarmonicMath. Cook, cyclist, Lost In Music. ∃🇮🇳 ∧ ∀🇬🇧 ∧ ∃🇪🇺 ∧ ∀🇺🇸 #celiac ex-{Microsoft, Google, Facebook, Xilinx}
Amit Shetty @amitshetty
2K Followers 4K Following Product @ Fox, Video|Ad tech|Ad quality|AI,ML. Prev- Product @ IAB TechLab,Yahoo!, Pixalate,etc. Tweets/views are my own…of course🤷 Blr-Aus-BayArea Longhorn🤘
Sam Buchanan @_sdbuchanan
2K Followers 1K Following prev. @Berkeley_EECS, @TTIC_Connect, PhD @EE_ColumbiaSEAS. Training efficiency, theory and practice!
Druv Pai @druv_pai
1K Followers 508 Following Research @thinkymachines, PhD @berkeley_ai | prev @GoogleResearch @GoogleDeepMind | science of scaling deep learning
Yi Ma @YiMaTweets
123K Followers 769 Following Chair Professor in AI, Hong Kong University. A Mathematical Theory of Intelligence/Memory: https://t.co/leZlkURb7j
Mitchell Gordon @MitchellAGordon
690 Followers 262 Following ML Engineer @Google. Views are to be abandoned.
Neha Verma @n_verma1
414 Followers 587 Following PhD student @jhu; Prev: research intern @Google+@GoogleDeepmind, @AIatMeta; @yale | Efficient models, merging, MT
Owen Oertell @owenoertell
2K Followers 228 Following research scientist @databricks. prev: bs @cornell reinforcement learning.
Ben Grimmer @prof_grimmer
3K Followers 443 Following Assistant Professor @JohnsHopkinsAMS, Optimization, PhD @Cornell_ORIE Mostly here to share pretty maths/3D prints, sometimes sharing my research
realfast @realfastai
2K Followers 35 Following Enterprise AI Transformations. Backed by @PeakXVPartners (formerly Sequoia India & SEA) and @RTPVC. We are hiring! https://t.co/3IugPQ6xcMYou might like
