We have another open call for a PhD position on ML system internals in my group: https://t.co/STUlW6OrjN (funded for up to 5 years, application deadline Nov 21)

Our 7th generation TPU (Ironwood) is becoming generally available for Cloud TPU customers. It's got greatly improved performance and efficiency over previous generations! 🎉

If anyone wants to know why Xiangpeng Hao is a great mentor, they can read this response: https://t.co/Zvu9c046DJ

The last two days, @BigEarthBIFOLD and I attended hearings at @PIK_Climate. We saw awesome candidates, great talks, interesting perspectives, and a beautiful campus. 🌳

2 #Postdoc positions at TU Wien (Vienna, Austria) 1️⃣ Data Warehousing & System Performance https://t.co/dVT8VlhXei 2️⃣ Natural Language Interfaces, LLMs, Exploratory Data Access https://t.co/bDuOyF9PCC Application Deadline: Nov 13, 2025 Details:

PyTorch announced Monarch which is meant to simplify distributed programming—your code looks and feels like a single-machine Python program, but can scale across thousands of GPUs. You can directly use Pythonic constructs—classes, functions, loops, tasks, futures—to express

New benchmark drop 🚀 SnorkelSpatial tests how well LLMs can think in space, following text-based moves and rotations in a 2D world.

Excited to release new repo: nanochat! (it's among the most unhinged I've written). Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,

Last week we had another FONDA II (collaborative research center) retreat, and here is a picture of Albert Einstein's summer house from a post-lunch walk.

How does @deepseek_ai Sparse Attention (DSA) work? It has 2 components: the Lightning Indexer and Sparse Multi-Latent Attention (MLA). The indexer keeps a small key cache of 128 per token (vs. 512 for MLA). It scores incoming queries. The top-2048 tokens to pass to Sparse MLA.

A postcard from ADBIS/TPDL 2025 in Tampere, Finland last week. It was a great and really enjoyable event.

Congratulations @SGrafberger on the successful PhD defense at @UvA_Amsterdam. 🎆 It was an extraordinary experience being on the committee and attending the event. All the best for the start at @Snowflake.

Dear Database Researchers, We are now seeking nominations from the database community for the 2025 Research Highlights. Check the following msg for details: https://t.co/jnjpCASAXW

We're releasing the DASLab GGUF Quantization Toolkit! 🚀 First open-source toolkit bringing GPTQ + EvoPress to @ggerganov's GGUF format, enabling heterogeneous quantization based on importance. Result: Better models at the same file size. [1/5]

A postcard from Berlin (some impressions from today's morning walk to the office)

To solve any problem, you don't have to be super smart. You just have to 1) be able to break down problems into subproblems, 2) be slightly smarter than the hardest of the atomic subproblems. The real challenge is that the process can take a very long time.

Got some hot research cooking? 🔥 The @satml_conf paper deadline is just 9 days away. We are looking forward to your work on security, privacy, and fairness in machine learning. 👉 https://t.co/cPFitltvjA ⏰ Sep 24

Teaching in the age of AI has brought challenges I never expected—even leading my students and I back to paper exams. 📄✏️ Fortune just published a thoughtful piece on how we’re navigating the role of humans and machines in education, research, and beyond. You can read it here:

Vol:18 No:12 → Enter the Warp: Fast and Adaptive Data Transfer with XDBC https://t.co/4xHs6FB2gg

Congrats to all distinguished AEs and reviewers.