What a future 🇹🇹 🇺🇸 MIT freshman looks like. Top of the class, Mandarin speaking, JavaScript coding, track & field ⭐️, and fashionista. Going to one of the top 10 middle schools in USA

BREAKING: NVIDIA just dropped an open 30B model that beats GPT-OSS and Qwen3-30B — and runs 2.2–3.3× faster Nemotron 3 Nano: • Up to 1M-token context • MoE: 31.6B total params, 3.6B active • Best-in-class performance for SWE-Bench • Open weights training recipe redistributable datasets You can run the model locally with 24GB RAM.

🚀 New blog: Building custom LLMs for public sector on AWS Learn how governments can develop national & domain-specific language models that meet sovereignty, compliance & cultural requirements. Full 6-stage development guide 👇
aws.amazon.com/blogs/publics… #AWS #AI #PublicSector #LLM#MachineLearning

had a sudden urge to write a neural network from scratch in C using MLX

this 17 year old homeschooled girl refuted a conjecture that was unsolved for 40 years, and which professional mathematicians worked on for years without solving she was rejected from most graduate programs she applied to, because she did not have a degree

Spent the last three days hacking DeepEP to work on EFA and I made it happen.

lets you build C/C apps that run on Linux, Mac, Windows, BSD, and BIOS

Learn how @nvidia Dynamo can be quickly setup & seamlessly deployed using Amazon EKS for automated scaling & simplified Kubernetes management ⚡💡🔧 NVIDIA Dynamo supports #AWS services such as Amazon S3, Amazon EFA & Amazon EKS. 👉 go.aws/3IK98YH

Grateful to be featured on AWS for AI Podcast (Ep. 8)! 🎙️ I shared insights on training foundation models at scale my journey from 🇹🇹 to AWS. Would love if you could watch, like & comment to support! 🙌 youtu.be/i95xUdpy0qQ?si=3wum…

First ever thermodynamic computer was put online internally today. Soon to be accessed by our first customers. So excited for things to come.

Today, we are releasing 4 hybrid reasoning models of sizes 70B, 109B MoE, 405B, 671B MoE under open license. These are some of the strongest LLMs in the world, and serve as a proof of concept for a novel AI paradigm - iterative self-improvement (AI systems improving themselves). The largest 671B MoE model is amongst the strongest open models in the world. It matches/exceeds the performance of the latest DeepSeek v3 and DeepSeek R1 models both, and approaches closed frontier models like o3 and Claude 4 Opus.

lock in. no one else's gonna do it.

Verification and Validation in Computational Science and Engineering by Patrick J. Roache.

Probability Theory and Mathematical Statistics:

I solved every single problem in the CUDA mode book. A quick thread summarizing this experience and what I learned 1/x

“The Coatue team mapped out cloud revenue market share, Oracle at 5%, Amazon at 44% with AWS, Google 19%, Microsoft 30%. Next to that, they showed NVIDIA GPU allocation: Microsoft and Google match their cloud share. Amazon is 44% cloud revenue but only 20% GPUs. Oracle jumps from 5% to 19%. Cori comes out of nowhere at 11%.” “One obvious takeaway is that Amazon has half the share of GPUs than their share of AWS. So that could mean one of two things, either AWS is behind in AI, that could be one, or they're pursuing a different hardware strategy than its competitors.” — Bill Gurley @bgurley and Thomas Laffont @thomas_coatue

🚨BREAKING: OPENAI IS RENTING GOOGLE’S TPUS TO HELP LOWER COSTS TO POWER CHATGPT

C 23 is not for toy-language coders. If you're playing with PHP, JS, Python, or Rust , stay in your lane. This is for real engineers with academic background.

The single most undervalued fact of linear algebra: matrices are graphs, and graphs are matrices. Encoding matrices as graphs is a cheat code, making complex behavior simple to study. Let me show you how!

AI is getting out of hand. The USA and China comp is serious and I think it’s good for AI because the gap between us and other countries is huge