The traffic sucks. Everything's expensive. But living in a place where convos abt physics, CS, and math are the daily norm rules.

Noveau getting these fixes is good. But it’s hard to see it gaining traction given the wide gulf between it and Nvidia official.

Mid: Illegally renaming DoD to look tough to Twitter nerds Actually good for the military: - Fixing black mold at Lejeune - Fixing bases like Hickam where service members literally drinking jet fuel - Paying NG actual deployment $$ when forcing them to do nothing in DC, LA

IMHO, the loss of Twitter/X as a platform where virtually everyone in infosec who was publicly active online had a presence has really been deleterious to a lot of info sharing. It's also no doubt prevented a lot of useful discussions and interactions from happening.

CoreWeave benchmarks the NVIDIA GB300 NVL72 📈 >6x higher throughput per GPU vs. H100 Read more about this breakthrough in our new blog → https://t.co/UzCAIk2WSO #GB300 #Deepseek #CoreWeave #AI

Get ready to rethink GPU cloud pricing. In the next episode of AI Cloud Horizons, CoreWeave's Urvashi Chowdhary and Harsh Banwait dive deep into TCO—beyond $/GPU‑hour. The full episode drops August 29, so reserve your spot here ➡️ https://t.co/s6IjF2KEtR #AICloud #TCO #CoreWeave

Been running Arch as my primary distro for last 5yrs and enjoyed it + its community. But with the continued DDoS, made the switch to Fedora. The AUR is an amazing accomplishment of OSS. But it needs much better resiliency given how critical it is to Arch https://t.co/2TjPABVgAk

“I think the major issue is inexperienced human drivers of AI that aren't able to adequately review their generated code.” This. This. Absolutely this. Nothing wrong with using AI generated code. Lots of issues with it being created blindly without experience.

CoreWeave engineers set out to test if a vertically integrated, GPU-first cloud can run large-language model training faster, cheaper, and more reliably than generalized hyperscalers. The results speak for themselves. Read the ungated whitepaper today → https://t.co/5hzJGQN1AA

It's a comparatively boring learning: it's not a superpower, it's not about being "smarter" or anything. But doing simple things like establishing a durable source of truth (e.g.: a Notion page for your functional area) and getting feedback on it is super valuable in ramping up

The most important thing that I've learned that's helped me ramp quickly as a product leader and manage a lot individually is organization. Beyond making sure I don't miss mtgs, sharing what I'm learning in an organized fashion helps get early feedback + build teams quicker

We’re proud to be ranked on the 2025 @Inc 5000 list of the fastest-growing companies in the U.S.! From powering cutting-edge AI to redefining cloud infrastructure, we’re helping the world’s top AI labs, hyperscalers, and enterprises innovate faster. https://t.co/QaT0Isiq39

Twitter *really* has gone downhill: - Ragebaiting vastly outnumbers thoughtful commentary to max engagement by bots for rev generation - Non-existent trust/safety team means lots of misinfo, even violent fascist content platformed

Last day at Kindo. Thank you for a great experience and the opportunity to work alongside a stellar eng team to move forward DevSecOps. If you’re at BlackHat, check out what we’ve built: https://t.co/UCny5vlpVz Excited about the big next adventure. But first some rest (and BF6)

Everyone's talking about AI agents as the next architectural shift. They are not a mystical, secret thing: they are just complex, long running, dynamic workflows. And engineers building them will run into familiar problems to anyone building distributed systems.

Nice little aide memoire table for PQC algorithms. https://t.co/cdhBTJcH1p

I’m moderating a fireside chat with @mitchellh at the @heavybit event on July 31st in SF. Make sure you register! We’ve got lots of stories to share (and maybe a few spicy takes). This one’s going to be fun. https://t.co/aKqxFyZVmx

We surveyed hundreds of engineers building in AI about everything from which models they’re using to whether they’re using a dedicated vector database. And, of course, if they think everyone will have AI girlfriends by 2030. Some highlights 🧵:

We should celebrate hacking, not cheating. Cheating quietly pockets extra Monopoly money (insider trading, plagiarism, inflated metrics). Hacking openly flips the game board, exposing rigged rules and inventing a better game for everyone: Wikipedia replaced Britannica, Airbnb

Kindo's WhiteRabbitNeo V3, a 7B Gen AI LLM for DevSecOps, impressed on Cybench CTF challenges—solving 5% of tasks autonomously, matching larger models like Llama 3 70B while running on just one A100 GPU vs 32+ for frontier models. Read about it here: https://t.co/8y1cohxmZm

In the end, LLM/LRMs may be a better exploration of the people in tech, than of the tech itself ... .. the greatest of all Rorschach tests.