Keep on to blur preview images; turn off to show them clearly
Should I swap TrustMRR homepage with the startup marketplace? - homepage becomes the marketplace - top 100 leaderboard gets a dedicated page
💻 https://t.co/Y30jsaHwz9 $30K/m ⚡️ https://t.co/vatLDmi9UG $21K/m 📈 https://t.co/3EDxln5mdi $17K/m ⭐️ https://t.co/MZc8tG9xWi $17K/m 🍜 https://t.co/r07EpGSYJ2 $1K/m 🧬 https://t.co/SfrVXVtmdA $0/m 🧾 https://t.co/7olaOzV8Xd $0/m +20 https://t.co/4zCWHGJp1S
You either die an early stage deep tech startup or survive long enough to suddenly have a $475M Seed competitor with the exact same thesis and blog post that is essentially your last round's pitch deck.
founder @ e/acc // building thermo compute @extropic // memetic warlord // Kardashev gradient climber
You either die an early stage deep tech startup or survive long enough to suddenly have a $475M Seed competitor with the exact same thesis and blog post that is essentially your last round's pitch deck.
Not implying any malfeasance here, just means the energy-efficient AI thesis is resonating. It's good that there are multiple bets in this sector, as this energy problem needs to be solved ASAP. Game on!
RT @MLflow: MLflow 3.7.0 is here and it brings major features and improvements for GenAI Observability, Evaluation, and Prompt Management!…
CTO at @Databricks and CS prof at @UCBerkeley. Working on data+AI, including @ApacheSpark, @DeltaLakeOSS, @MLflow, @DSPyOSS. https://t.co/nmRYAKFsWr
Code RED for India 🚨 - only paranoids survive. ============ India is nowhere in Neurips 2025 in San Diego. World's largest AI research conference. Below is leaderboard of sorrow for India. India Universities are nowhere to be seen. India's dream of becoming a technological power by 2047 will be a dream only, as there is no serious action on the ground. @OfficeDp, @AshwiniVaishnaw, @PMOIndia - you have the most power to change it. You have power for more than 11 years. That is a lot of time, I would say given how fast the world changes. Either nothing has been done that matters, or it is not enough. Did meet a few India based researcher, but they are so few to spot. Indias do well in AI outside India, not in India. And it has to be a matter of national shame. As decision makers - learn from the best: ---------------- Look at the performance of Chinese Universities - INCREDIBLE Look at contribution of Singapore universities (NTU, NUS at bottom in Pink). Introspect what India is doing wrong. whatever you are doing is not clearly working. What can India learn from China that has caught up so rapidly? Just copy that. Your advisors are not that smart. 180 IQ babus are ill qualified. IIT are highly overrated...let us stop celebrating them ---------------- Students are great (after all they selected to be 0.1%). - they do well. The institutions are lacking, however. If there is not emergency meeting of leaders of all IITs after this to introspect, probably you are not doing your job right. India has lot of smart people, and retirement is not all that bad. It is not only about GPUs..incentives matter too: ---------------- Most student papers are published after conducting research on servers with 8 GPUs. This research prepares students for research on larger clusters (it is the first step). Universities need GPUs, but it is also a matter of incentives. There needs to be strong mandate for IITs and other universities to publish paper. Research is a social activity: ---------------- Unlike what most people in India believe, researchers don't work in isolation. They know other researchers across the world working on same problems through social media (X, Reddit) or conference events. Unless Indian students and professors become a part of international research circle, this is not going to change. Research is a social activity. India needs to to spend a few billion dollars every year to send thousands of student (it can cost 200K INR to attend one such conference). 1B USD can suffice to send 500K students to such conferences annually. It is loose change for a country research 5 trillion in GDP size. Not sorry for the harsh words ---------------- Headline management is not enough. You are paid well and you are celebrated. And you have definitely done a few great things (Semiconductors initiatives, Railway Modernisation, AI DCs etc). I am aware of some initiatives by likes of IIT Madras and some funding for AI startup. Scores are out and clearly that is not enough....at least not enough in 11 years. I know you are smart and can fix this if you put your mind to it, but yeah I will reverse my opinion when I see some change that shows India is preparing next generation of AI researchers in serious numbers. What should be the goal for India?: ---------------- Most of these papers are published by very young researchers (below 20 or early 20s). You don't need 20 years of experience to be at the frontier of deep learning (technology that powers AI). Most of them are self taught using ChatGPT and Claude. What they have is mandate from their universities for this type of work and GPUs. 16% of all the people on this planet are Indians. One should expect Indian Universities to have 800 papers, if the conference accepted 5000 papers (like this Neurips conference did). I don't think there were even 1% from Indian Universities. Don't give isolated examples of brilliance. Token activities are not appreciated. Don't care for the consolation prize. If Education Ministry and University leadership can't improve these numbers year on year, may be time is to retire and pass control to younger people. Start by copying - that is what Eklavya did ---------------- Thanks to lot of great work by likes of @huggingface , @allen_ai , @PrimeIntellect and Chinese foundation model labs lot of knowledge is available in open source/open domain. Students should replay the same training and post training pipelines first and then start tinkering with them. Learn how evaluations are built, then build your own evaluation. Tear apart inferencing engines to see how they work. Ask ChatGPT 5 to explain you how the Reinforcement Learning math works. Doesn't matter if your professor doesn't know these things. You copy the masters across the world - I will be writing more on this in coming days. Corporate leaders your generosity is needed: ---------------- @gautam_adani, Mukesh Bhai, @anandmahindra your generosity is needed. Leading Indian universities need at least 1 GPU per student (at least for Comp Science & Electrical Engineering students). Regarding people tagged: ---------------- I know most people tagged in this thread are not going to react or reply, as my words are harsh, while being respectful. I will be grateful if they did. But, yeah they would be rightfully worried of repercussions. India clearly has culture of sycophancy & self congratulation. ---------------- Direct feedback is not given or appreciated. In China, despite not having the kind of democracy India has; objective feedback is very much appreciated and respected. There are lively and data driven debates within CCP. India could learn a bit from that. This culture of sycophancy & self congratulation will prevent India from this 2047 ambition. Forget it, if India can not introspect without getting feelings hurt. Need to grow up as a nation. (these opinions are mine and mine only). @vikramchandra @chandrarsrikant @AbhijitChavda @TheJaggi @svembu @HarveenChadha @vijayshekhar @ShekharGupta @bhash @deedydas @Iyervval @cneuralnetwork @RajeevRC_X @RMantri @hvgoenka @harshmadhusudan @MohapatraHemant @TVMohandasPai @HindolSengupta @dhume @vikrantyagnick
Artificial Intelligence @amazon, @awscloud Reinforcement Learning, OSS AI, Coding Agents, General Purpose Agents All views personal - I only represent myself!
RT @EBorgnia: Introducing Relace Search, a utility agent designed for fast codebase retrieval. RAG is fast because it's fully parallelizab…
Cartoonist, Engineer, PM, Partner @a16z investing in infra & AI Prev Product lead @HashiCorp, Founding Eng/PM @Transposit. Eng @AppDynamics. Opinions = own.