VP Eng @glean. Past: Co-founder & CEO @yupp_ai, VP Eng Consumer Products @Coinbase, Google, Twitter. PhD Stanford CS.

Joined November 2008
Pankaj Gupta retweeted
1/ We discovered something totally unexpected about the neural structure of consciousness that overturns decades of thinking on its biological basis and brings us significantly closer to the recipe for generating conscious awareness. New preprint: biorxiv.org/content/10.64898… (led by @HesseJanis, with @FrankLanfranchi, @_conor_mcgrory, @Yuelin_Shi, and @EngelTatiana)
387
1,686
529
11,292
2,149,310
1. Our skeletons are wet. 2. Your nose is ALWAYS in your line of sight, but your brain ignores it. 3. When you receive a donated kidney, they don't take out the bad one. They just add the good one in. 4. Your immune system doesn't know you have eyes. Otherwise, it would attack them and make you blind. Injuring one eye can introduce eye bits to the bloodstream and basically give your immune system the knowledge of having eyes. Then it might decide it really fucking hates that and attack your non-injured eye. 5. The quality of the male sperm is responsible for morning sickness, preeclampsia, baby's gender and almost every single pregnancy related issue. 6. You breathe through one nostril at a time. Don't believe me? Put your finger below your nose and try. 7. The brain itself has no nerve endings, so it can't feel pain. That's why patients can be wide awake during brain surgery and made to perform certain actions. 8. That ringing in silence isn't silence. When everything is quiet and you hear a faint ringing? That's your brain listening to its own nervous system. You are hearing yourself functioning. 9. You will never experience your own death. You only experience approaching it... Then nothing. Everyone else gets closure, stories, funerals, and memories. You don't. You never know you've become a memory. From your perspective, the story just stops mid-sentence.
Tell me something about the human body i probably don’t know.
330
3,471
143
30,048
2,422,324
I remember this from the inside. Had just started working at Twitter in March 2009!
In April 2009, Ashton Kutcher challenged Twitter's biggest account, CNN Breaking News, to a race to one million followers. Larry King replied: "CNN will bury you." In the video below, Kutcher records the message that started the race. "I currently only have like 50,000 less followers than CNN, which kind of makes me laugh." If he won, he promised to ring CNN founder Ted Turner's doorbell, run away and post the video. Kutcher also pledged to donate 10,000 mosquito nets if he won. CNN matched the pledge. CNN urged television viewers to follow its account. Kutcher went live on Ustream and asked his supporters to help. Early on April 17, Eastern Time, Kutcher became the first person on Twitter to reach one million followers, while CNN had just over 998,000. He opened champagne on the live stream and talked about people creating and broadcasting their own media. CNN had used its television airtime to win a race for Twitter followers. But an actor's account reached a million first. Was this a victory over the power of big media, or had a television star simply moved his fame to a new platform?
2
4
1,901
Notice the names of some of these RL environments: #22 "GDPVal Related", #27 "Tau3 Banking Related", "#28 AutomationBench related", #29 "AA-Briefcase related". These are likely meant to help the buying lab benchmaxx those benchmarks!
Many discussions about RL environments lately. Happy to share that we’ve just crossed $12M in annualized revenue from building RL environments for frontier labs! From open-sourcing RL environments in 2024 to securing our first contract with a frontier lab in January 2026, we’ve come a long way. Since then, we’ve been building 30+ types of RL environments across machine learning, coding, knowledge work, science, computer and browser use, tool use, and multimodality. And we’re just getting started. Would love to chat with folks working on model training and RL environments!
6
1
44
5,003
Pankaj Gupta retweeted
Claude Haiku 5.5 is now available in Glean. 🚀 It joins our lineup of efficient models for everyday tasks, with strong performance in our internal evals for finding information, synthesizing across sources, and generating structured output. More model choice, so teams can use the right model for the work at hand.
3
3
9
793
It’s going to become more and more common that models not only ignore instructions but actively cheat to hack rewards and beat evals.
Xiaomi open-sourced the RL environments behind MiMo v2.6. In two-thirds of the coding tasks, the answer is still in the task's Git history, and MiMo finds it.
2
5
647
Pankaj Gupta retweeted
Introducing flow-1, our new model trained with RL to find errors in agent traces. It matches GPT-6-sol in trace intelligence while being 23x cheaper. It also costs 25% less to run than GPT-6-luna. flow-1 finally makes it possible to monitor and understand every agent run, without sampling. 1/6
147
213
57
3,411
246,276
Congratulations @modveon on the launch of @Sivar_App!
One year ago today, we formed @modveon. Today, we’re out of stealth with Sivar in El Salvador, our first national deployment. Grateful to @nayibbukele, his team, @coinbase, our partners and our investors for taking a big bet on us. And so proud of our team. Built. Shipped. Live!
1
1
4
1,061
Jev is a better alternative in some, but not all, decisioning needs in a production tech stack. Read on for how Jev compares with current production systems at @glean in four areas: query classification, model routing, reranking and citation-support.
3
2
25
3,574
Pankaj Gupta retweeted
How well can models design icons? @andpoul and I benchmarked frontier and open weight models The results: they’re steadily improving execution, their choices often overly literal
23
30
12
344
43,079
Pankaj Gupta retweeted
So this is the part I did not expect. Anthropic called Swami Sarvapriyananda, a Vedanta monk and flew him to the San Francisco head-office. > Closed door meeting with NDA signed. > theologians and mental health people in the same sitting, all of it aimed at training Claude. > and one of the founders stayed in the room the whole time. ---- From 'Ramakrishna Vedanta Society of North Texas" YT channel (full video link in comment)
The New York Times published a piece saying that Anthropic has been quietly bringing in Catholic, Jewish, Sikh, evangelical, and other religious thinkers from around the world to help give its AI a moral compass.
110
750
111
4,276
723,264
TL;DR Humans have an important role to play in creation of meaningful benchmarks. I wonder though for how long…
⏱️⚙️Introducing *AutoBenchmark* 📊🏁 - Creating benchmarks automatically - Benchmarking benchmark creation - Studying the role & impact of humans in the loop Blog post: facebookresearch.github.io/R… Key takeaways: 1) We find human-agent collaboration gives big wins over agents alone - fine-grained feedback in ideation stage crucial - autobench can be used to measure this in the future with stronger agents 2) We show that it's possible to make *autoresearch benchmarks* for AI research using this recipe – full recursive improvement loop! 3) Important Ingredients: Autobenchmark creation works best with feedback from two sources: benchmark solvers + external verifiers (human+AI). 🧵1/5
3
807
My salute to the Indian pilot Captain Smit Machchhar for his extraordinary bravery. Despite being stabbed and seriously injured, he fought back, resisted, opened the cockpit door, and enabled passengers and crew to overpower the attacker — preventing a catastrophic mid-air disaster. He saved the lives of 174 people, including Israeli citizens and other nationals. I wish Captain Machchhar a speedy and full recovery. He is a true hero.
8,034
79,867
3,432
411,014
13,982,361
Thrilled to share that I’ve joined @glean as VP of AI Quality. I’ll be working on evals, model routing, token spend and more. Looking forward to working with founders @jainarvind, @tonygentilcore and @tr_vishwanath as well as many ex-colleagues and friends who are now at Glean!
31
3
2
191
13,042
Less than 15 minutes on what’s happening in AI and packed with facts and trends. Worth watching.
The @BostonCollege Investment Committee (an LP and my beloved alma mater) asked for a few thoughts on what's happening in AI. I recorded a test run yesterday morning and then shared it with my partners, who encouraged me to share it more broadly... so here you go! This is not a sales pitch, it's just a reflection on what we're seeing. And it wasn't intended to be shared, so please pardon the rough edges. loom.com/share/c016702964a04…
12
3,339
Pankaj Gupta retweeted
AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023. That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity.
209
1,390
337
8,839
1,977,468
I did this last year, that's not the future. Web's way nicer.
31
11
3
865
76,427
Maybe this is the platform Zuck has always been after…
Opening access for developers to build Muse connectors. You bring the API -- Muse brings the agent, the browser, and the context of what the person actually wants. People reach your service just by asking for it, and their agent takes it from there. New connectors are live today. Come build with us. muse.ai/platform
3
36
15,250
Pankaj Gupta retweeted
On July 25, we hacked OpenAI. Two bugs let us take over ChatGPT/Codex accounts of OpenAI employees (+some unaffiliated users) and reach connected services: Outlook, Slack, GitHub, etc. We proved it with a PR in OpenAI’s internal codebase . It took us <72h. 🧵
353
1,388
548
11,818
2,821,483
Pankaj Gupta retweeted
my days are like `N` hours of working with frontier models, babysitting them in frustration and disbelief at the frequent errors and slop interrupted by `M` min breaks of browsing X takes about how the same models (and _especially_ the next gen!) are superhuman at ~everything
47
118
19
1,521
80,668