🔥Telegram İfşa

Hey Department of Government Efficiency gorklon rust, I can save taxpayers $900M+ annually. This is not a joke. Let's talk. The federal government spends over $8B/year on AWS and other clouds. This is one of the biggest budget line items. Shockingly (or not), little of it is tracked and most is idle and

AI is starting to design new antibiotics to fight hard-to-treat diseases, showing real promise. But the risks remain clear—when doctors relied too heavily on AI, their detection skills slipped, and even dangerous suggestions caused harm. The AI hype is real, but so are the

The uncanny valley is officially closing. Synthesia's new Express avatars can now replicate human emotions and movements with unprecedented realism, marking a pivotal moment in AI generated content. What makes this breakthrough significant: These avatars learn from just 10

Check out the community I created: x.com/i/communities/…

Need suggestions on how to generate synthetic lung cancer datasets.

Ever wondered how much juice those beastly AI training rigs really suck up? I came across this fresh IEEE paper measuring power on an 8-GPU NVIDIA H100 node—max draw hit just 8.4kW (18% under the 10.2kW rating!), even at near-full throttle. Training Llama2-13b? Median 7.9kW with

ieeexplore.ieee.org/abstract/docum… This research paper measures the power used by a computer system with eight NVIDIA H100 GPUs during artificial intelligence training. The system is a single node, which means one unit in a larger setup like a data center. The researchers trained two

Don’t say anything, just leave a thumbs up 👍 that's all he needs from you

Help me Choose?

Comparative research on open-source CI/CD tools for ML shows distinct tradeoffs. Jenkins offers flexibility but with a steep learning curve, GitHub Actions excels in usability but struggles at scale, and Bitbucket Pipelines stands out for resource-intensive tasks with Kubernetes

"Introduction to Machine Learning Systems" - FREE from MIT Press - Authored by Harvard Professor - 2048 Pages To Get It Simply: 1. Retweet & Reply "ML" 2. Follow so that I will DM you.

ML

One thing that I am constantly learning when building a startup is that you have to have a great team. People should believe in you and your vision. My experience to date is people show fake commitments that they will work and after 2-3 days they stop responding.

Seems like Anthropic really took the Tylenol with this one

Lets try Claude Sonnet 4.5 and its capabilities.

claude 4.5 is really solid Anthropic has done a great work. there was a bug which was not getting solved since 4 hours and since the update claude code has this amazing understanding of what the user wants. It solved it in 5 minutes by pin pointing where the error was.

. Anthropic you should add this rollback to the previous version feature I have struggled a lot when claude made a mistake and i wanted to rollback to the previous version and it was a mess. I really want this feature. Still testing 4.5 sonnet so far so good.

Just updated ios to ios26 did you all updated it? If yes did you liked the Liquid Glass update? The animations are more smoother too. It happens with me like i really don’t like to update the OS because I am too habitual of the old one. They have worked in the keyboard but it

A Microsoft team used AI to find a new weakness in biosecurity systems that screen DNA orders to prevent the creation of dangerous toxins or pathogens. Their research, published in Science, showed how AI can redesign toxic proteins to bypass these safeguards while keeping their

We are in the residency