Mahesh Kumar

I’m excited to share that I will be joining @demishassabis and @GoogleDeepMind as Chief Strategy Officer! I feel a moral obligation to shepherd AGI into the world responsibly, benefiting everyone. I am looking forward to doing this with the team at DeepMind.

Unveiling our new startup Advanced Machine Intelligence (AMI Labs). We just completed our seed round: $1.03B / 890M€, one the largest seeds ever, probably the largest for a European company. We're hiring! [the background image is the Veil Nebula - a picture I took from my backyard, most appropriate for an unveiling] More details here: techcrunch.com/2026/03/09/yan…

really good analysis of what is actually your 'moat' Loved the muscles for strength analogy @sidrmsh

@hasantoxr @grok when was LangWatch first available and what is the delta of features since then

🚨 BREAKING: Someone just open sourced the missing layer for AI agents and it's genuinely insane. It's called LangWatch. The complete platform for LLM evaluation and AI agent testing trace, evaluate, simulate, and monitor your agents end-to-end before a single user sees them. Here's what you actually get: → End-to-end agent simulations - run full-stack scenarios (tools, state, user simulator, judge) and pinpoint exactly where your agent breaks, decision by decision → Closed eval loop - Trace → Dataset → Evaluate → Optimize prompts → Re-test. Zero glue code, zero tool sprawl → Optimization Studio - iterate on prompts and models with real eval data backing every change → Annotations & queues - let domain experts label edge cases, catch failures your evals miss → GitHub integration - prompt versions live in Git, linked directly to traces Here's the wild part: It's OpenTelemetry-native. Framework-agnostic. Works with LangChain, LangGraph, CrewAI, Vercel AI SDK, Mastra, Google ADK. Model-agnostic too OpenAI, Anthropic, Azure, AWS, Groq, Ollama. Most teams shipping AI agents have zero regression testing. No simulations. No systematic eval loop. They find out their agent broke when a user tweets about it. LangWatch fixes that. One docker compose command to self-host. Full MCP support for Claude Desktop. ISO 27001 certified. 100% Open Source. (Link in the comments)

@iruletheworldmo Context seems to be everything. Even in X posts - where readers add context 🤣

gpt 5.3 drops tomorrow and we’re starting to see early testers leaking results. huge wins across the board as they brutally framemog google and anthropic. hearing this will be the first time we get a big model from openai since 4.5 sama finally getting his compute online. one insider mentioned this run is ‘our first taste of hardmaxxing’ from a major lab, though i’m not sure what this means.

@SHEKARSUSHEEL Oh boy! This brought back some serious memories - Guran, Ms. Tagama, Diana, Rex, Hero, and so on...

The Phantom turns ninety today. Time passes and generations change, but the Ghost Who Walks remains eternal. Celebrating his ninety years feels like celebrating a part of my own journey, the dreams, the inspiration, and the quiet strength he planted within me long ago.

While I get the argument of asset vs. shelf, and that it is easy to whip up code/software, for an ent (not ISV) is the argument that we will see bespoke CRMs (pick your SaaS poison and replace) (btw we have built one), sprouting up everywhere. I do feel the friction between vibe coding a not very useful CRM pilot/demo/prototype and one that will meet a Fortune 100 company's needs is not insignificant. For sure SaaS amortization is questionable, seat licenses and expansions and add-on modules all are under the microscope. But gut tells me ISVs will also adapt and we will likely reach a hybrid medium that leverages ISV intelligence built over the years to deliver (overloaded term) almost bespoke, custom software at far lower margins. Maybe....

Thanks for hosting me @iiscbangalore! Really enjoyed talking with Prof. Rangarajan & Varun Mayya @waitin4agi_ about AI for Science. Impressed by the energy and enthusiasm for AI in India, especially from the young. Great to see the statue of Ramanujan, one of my all-time heroes!

@vkhosla Instead of all of nothing even starting to eliminate for the 40% making >10M would be a good start to see the effects without adversely penalizing everyone’s capital gains?

AI will transform economies and need a rethink of capitalism & equity. Labor portion of economy (vs capital) will decline sharply. Should we eliminate preferential treatment of capital gains tax and equalize to ordinary income? 40% of capital gains taxes are paid by those with income >$10m/year! This video is just the professions being automated in our current portfolio!

@elonmusk @elonmusk tongue in cheek - what year sir? 🤣

Cybercab, which has no pedals or steering wheel, starts production in April

I've been waiting 3 years to show you this We just launched our 3rd-gen humanoid, but we’re already on our 7th-gen hand Our team has quietly worked for years to approach parity with a human hand Excited to share a sneak peek of some of the best engineering I’ve ever seen

GPT-5.2 derived a novel result in theoretical physics, showing that a type of particle interaction many physicists expected would not occur can in fact arise under specific conditions. There is great promise in the potential of AI to benefit people by accelerating science.

Hello World! - from team “Tenkasi Semiconductors” Stay tuned….

@rameshsrivats Missing RDB

Very nicely done

Congrats to our hometown Super Bowl champs @Seahawks! So thrilled for our city and the 12s!

Seattle Seahawks 🏈

@ylecun's vision: machines should learn like humans — by building internal world models, not reconstructing every pixel. We just validated this idea at the largest scale ever attempted in cardiac ultrasound. Introducing EchoJEPA — the first world model for medical video. 🫀 18M echocardiograms 👥 300K patients 🧠 Learns heart dynamics — not imaging noise The problem: Ultrasound is messy. Speckle, shadows, attenuation. Most pretraining objectives end up modeling the scanner, not the heart. The idea: Stop reconstructing pixels. Predict latent structure instead. EchoJEPA discards what’s unpredictable and locks onto what matters clinically: ➡️ chamber geometry ➡️ wall motion ➡️ valve dynamics The results (frozen encoder, no fine-tuning): • 20% ↓ error in LVEF • 17% ↓ error in RVSP • 79% accuracy with 1% labels (vs 42% for baselines w/ 100%) • 2% degradation under acoustic artifacts (vs 17%) • Zero-shot pediatric transfer beats all fine-tuned models Why this works: When we project embeddings: ❌ prior methods → diffuse, entangled clusters ✅ EchoJEPA → clean anatomical organization Structure separated from acquisition noise. 📄 Paper: arxiv.org/abs/2602.02603 💻 Code: github.com/bowang-lab/Ech… Huge credit to @alifmunim , who pushed JEPA thinking into medical video and led this effort 💥 Guidance from @AIatMeta (@garridoq_, @koustuvsinha) Co-authors: @adibvafa @TeodoraSzasz @A_Attarpour @riverjiang @JSlivnickMD Teams: @UHN @awscloud @UofT @UChicago @UCSF @PhilipsHealth This is representation learning for physiology, not pixels.

I’d like to hire strong data engineers to join our Microsoft Super Intelligence (MSI) team. I am interested in people who are good at processing PDFs and other documents at billion scale, and people good at parsing the web at trillion scale. If you dream of processing all of human knowledge to advance science and engineering, this is for you. Also looking for strong evaluation and post-training engineers. Be part of our first launches this year 🚀 We have all the resources in the world to support you, working in startup mode, while powering a large organisation with billions of users. Hiring in London, Zurich, New York, Boston, Toronto, Seattle and SF. Please send your CV to JoinAITeam@microsoft.com

The companies that succeed in the future are going to make very heavy use of AI. People will manage teams of agents to do very complex things. Today we are launching Frontier, a new platform to enable these companies.

@elonmusk lol the advise pours in while answer lies within @elonmusk 🤣

Whoever said “money can’t buy happiness” really knew what they were talking about 😔