Mahmoud Soliman

Join the JAX team at Nvidia, If you want to contribute to state of the art GenAI models or nobel prize winning science advances! Check out this job at NVIDIA: linkedin.com/jobs/view/3663…

🚀 MLSys 2026 Contest - @nvidia Track is LIVE! Registration is now open for the FlashInfer-Bench Challenge! Submit high-performance GPU kernels for cutting-edge LLM architectures on NVIDIA Blackwell GPUs. Three Tracks * MoE (Mixture of Experts) * DSA (Deepseek Sparse Attention) * GDN (Gated Delta Net) Human experts AND AI agents welcome — evaluated separately. Let's see who builds the best kernels! 🤖 🎁 Prizes: Winners take home NVIDIA GPUs and are invited for presentation at MLSys 2026. ⚡ First 50 teams to register get free GPU credits from @modal - huge thanks for the sponsorship @charles_irl ! Whether you're a kernel wizard or building autonomous coding agents, we want to see what you've got. 🔗 Contest details: mlsys26.flashinfer.ai See you at MLSys 2026! 🔥

My team is looking for interns! Please reach out with your resume if you want to contribute to the state of art deep learning frameworks and compilers, powering frontier models on pre/post training and inference!

@tenderizzation We have only “Bounded” dynamic shapes, sir.

@tenderizzation The problem is you didn’t pad your input (user error)

“why yes just give us your model and our state-of-the-art compiler will automatically shard it across your cluster”

That’s for the DGX Spark 😀 This is ~100X more compute per watt than the DGX-1, the first ever dedicated AI computer, that Jensen gave me at OpenAI in 2016!

NVIDIA has widened the gap for inference rack scale architecture yet again! Prefill specialized inference chips massively lower TCO per million input tokens on long context transformers As usual, other AI chip upstarts will follow this with prefill specialized chips, but later

@ahmadalfy @ahmd3ssam I believe the views w.r.t software are 2012 , the 2025 picture is totally different… if you look at GitHub and look at how many containers that are Linux optimized, those paint a different picture. Food for thought. All GenAI train and mostly inferenced on Linux for example.

@ahmd3ssam غير التاريخ السئ بتاعها تجاه ال open standards … يعني الواقعة الشهيرة ل Linus في سنة ٢٠١٢ لما رفع ال middle finger وقال دي أسوأ حاجة احنا بنضطر نتعامل معاها. معتقدش أن اي حاجة من الحاجات دي اتغيرت تعالى بقى في وقت COVID وازمة ال supply chain بتاعت ال chipsets …

الشركة في طريقها للـ١٠ تريليون تخيل ان كل روبوت لازم هيستعملوا فيه تشيب انفيديا انفيديا لو قررت متعملش حاجة لمدة ٥ سنين هيفضلوا برضة بيكسبوا قبل ما تقول اي حاجة بص على الـecosystem بتاعهم الاول

@ahmd3ssam @ahmadalfy developer.nvidia.com/blog/nvfp4-tra… we are just getting started 🔥

@ahmadalfy ليه؟ حاجة ليها علاقة بالألعاب ؟ و لا علشان ماديين و ميعرفوش ولاد خالتهم 😂

@GaryLHenderson @eightsleep @JNAranda I was able to cancel when I called them directly. Their CS when faced with “Your website is quoting a delivery date earlier than what you said my delivery would be” stated that the website dates can be delayed too. I don’t know what to call this!

@GaryLHenderson @eightsleep Order made on June 15th and got pushed twice out to beyond what a new order date is! AI loop is a not an excuse, they should have a report that shows order fulfillment rate. @JNAranda another cancellation here please.

Ordered a Pod 5 from @eightsleep over two weeks ago and here's my experience. 1 - No proper order confirmation or communication. 2 - After 15+ failed attempts at email support a human finally responded and told me I got trapped in their AI and apologized. They told me it would ship and it didn.t 3 - After multiple follow ups and request to escalate to a supervisor I kept getting ignored 4 - I called and asked to speak with a supervisor and was told none was available. 5 - I asked to cancel my order and was asked my order number. 6 - I provided and they had my order date and product wrong. 7 - I asked for an immediate cancellation and refund and was placed on hold for multiple minutes. I would never spend a dime with this company and I suggest you don't either if you value your time and money.

@0xA95 @seanprime7 @vinodg ‘s work is finally out. Kick the tires and let them know what do you think!

@bryan_johnson And at 2.0 ATA the MacBook didn’t crack?

Moved my office into my hyperbaric oxygen chamber.

Nemotron-H: A family of Hybrid Mamba-Transformer LLMs. * Hybrid architecture means up to 3X faster at the same accuracy * Trained in FP8 * Great for VLMs * Weights and instruct versions to come soon. research.nvidia.com/labs/adlr/nemo…

@grok is so good! Congrats @ibab and the team for going from zero to the front so fast!! Even the iOS app is great. The competition is heating up fast!

#ReviewerIIStrikesAgain !

This month is my last at Google. In a month, I'm joining Google DeepMind. Our job will be making DeepMind's work publicly available in ways that wasn't possible before. While this is one of the biggest challenges in my life, it's potentially going to be the most rewarding.

Introducing 🧞Genie 2 🧞 - our most capable large-scale foundation world model, which can generate a diverse array of consistent worlds, playable for up to a minute. We believe Genie 2 could unlock the next wave of capabilities for embodied agents 🧠.

The #JAX team at @nvidia is looking for interns for Summer 2025 nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAEx…

Excited to share that the AlphaFold 3 model code and weights are now available for academic use. Looking forward to seeing what new research this unlocks and how the research community builds on AlphaFold 3 for scientific discoveries github.com/google-deepmin… 1/2