K-Means is simple. Making it fast on GPU isn't.... Flash-KMeans is an IO-aware implementation of exact k-means that rethinks the algorithm around modern GPU bottlenecks. By attacking the memory bottlenecks directly, Flash-KMeans achieves: - 30x speedup over cuML - 200x speedup over FAISS Using the same exact algorithm, just engineered for today’s hardware. At the million-scale, Flash-KMeans can complete a k-means iteration in milliseconds. Here's why this matters today: K-means has always been an offline primitive. Something you run once to preprocess data and move on. These speedups change that. ↳ Vector databases like FAISS use k-means to build search indices. Faster k-means means you can re-index dynamically as data changes, not batch it overnight. ↳ LLM quantization methods need k-means to find optimal weight codebooks, per layer, repeatedly. What takes hours could now take minutes. ↳ MoE models need fast token routing at inference time. Millisecond k-means makes it viable to run this inside the inference loop, not just in preprocessing. The 200x over FAISS is the number to internalize. FAISS is the industry standard. Most production vector search systems sit on top of it. Link to the paper and code in next tweet!show more

Its not every day you wake up to find... that the Pope has made your lifes work the central focus of his papacy: “Disarming AI means freeing it from the mentality of “armed” competition [..] This entails a race for ever more powerful algorithms and larger datasets, driven by the desire to secure geopolitical or commercial dominance." - POPE LEO XIV, May 2026 Here, “disarmed” means “neutralised” in the sense that this should not be a differentiator The Innovation Game (TIG) was created to keep data and algorithms open, in order to prevent monopolistic control It's not just an aspiration, it's an economic mechanism that makes open data and open algorithms the rational economic choice • All algorithms are published openly by TIG • If you are willing to make the data you process with an algorithm open, you can use it free-of-charge • Alternatively, if you would like to keep this data private, there is a fee to pay for using the algorithm • All fees are used to fund more open innovation Its an elegant, global, self-reinforcing engine The logical end point of monopoly is that innovation stops We cannot allow that to happen Pope Leo XIV I would be grateful for your thoughts on The Innovation Gameshow more

Big moment for Postgres! Search has always been Postgres'... weak spot, and everyone just accepted it. If you needed a real relevance-ranked keyword search, the default answer was to spin up Elasticsearch or add Algolia and deal with the data sync headaches forever. The problem isn't that Postgres can't do text search. It can. But the built-in `ts_rank` function uses a basic term frequency algorithm that doesn't come close to what modern search engines deliver. So teams end up: - Running a separate Elasticsearch cluster just for search - Building sync pipelines that inevitably drift out of consistency - Paying for managed search services that charge per query - Accepting mediocre search relevance because "good enough" ships faster But this is actually a solvable problem. You can realistically bring industry-standard search ranking directly into Postgres, which eliminates the need for external infra entirely. This exact solution is now available with the newly open-sourced pg_textsearch by Tiger Data - Creators of TimescaleDB, a Postgres extension that brings true BM25 relevance ranking into the database. BM25 is the algorithm behind Elasticsearch, Lucene, and most modern search engines. Now it runs natively in Postgres. Here's what pg_textsearch enables: - True BM25 ranking with configurable parameters (the same algorithm powering production search systems) - Simple SQL syntax: `ORDER BY content 'search terms'` - Works with Postgres text search configurations for multiple languages - Pairs naturally with pgvector for hybrid keyword + semantic search That last point matters a lot for RAG apps. The video below shows this in action, and I worked with the team to put this together. You can now do hybrid retrieval (combining keyword matching with vector similarity) in a single database, without stitching together multiple systems. The syntax is clean enough that you can add relevance-ranked search to existing queries in minutes. pg_textsearch is fully open-source under the PostgreSQL license. You can find a link to their GitHub repo in the next tweet.show more

OpenAI's Deep Research is getting a run for its... money. Deep Lake was just released, and it's a different take on an AI system that can do deep research on your own data. You can use Deep Lake to build AI search with reasoning on your private and public data. (Look at the attached videos to get an idea of how it works.) If you want to research proprietary and sensitive data, Deep Research won't help you because it's limited to public data. Deep Lake, however, will allow you to use your private data. On top of that, Deep Lake supports multi-modal retrieval from the ground up. It uses vision language models for data ingestion and retrieval so that you can connect any data (PDFs, images, videos, structured data, etc.) You can even use mixed-data queries! Deep Lake can search your data from S3, Dropbox, and GCP. It learns from your queries over time, making the results as relevant to your work as possible!show more

forget the $699 AI pins. this $8 chip just... shattered the barrier for local AI hardware. a developer just forced a 28.9 million-parameter LLM onto a standard ESP32-S3 microcontroller. it costs roughly 8 dollars, runs completely offline, and draws the power of a single LED. conventional wisdom said a model of this size simply would not fit. the chip only has 512 KB of fast SRAM and 16 MB of flash. the breakthrough is architectural. the developer moved the bulk of the embedding table into flash memory and memory-mapped it. the chip only needs to pull about 450 bytes per token, keeping the active working memory inside the fast SRAM. this means you can now embed a capable language model into a physical node for the price of two coffees. and we are already seeing the beginnings of this custom physical hardware. in the video, a creator built a minimalist voice-controlled universal remote using an ESP32. it captures voice and remotely controls the computer over bluetooth LE. he simply says "open chrome and open 20 new tabs", and the custom hardware executes it instantly. we have spent years watching model sizes explode upward. but the true frontier is the opposite direction. when an eight-dollar chip can power offline intelligence and custom physical interfaces, AI becomes local infrastructure rather than a cloud service.show more

AI token usage is up 10x in 7 months,... compounding 40%/MONTH! There is NO BUBBLE when demand is STILL accelerating And this is just OpenRouter, it doesn't count the labs direct token usage and APIs But here's what's interesting about these numbers, the demand is coming from everywhere at once US models (OpenAI, Anthropic, Google) keep growing, while Chinese open weight models (DeepSeek, Tencent, Xiaomi, Minimax) grew even faster and now drive over 60% of usage on OpenRouter Closed source and open source both compounding at the same time. This is literally the best case scenario for AI Infra investors It means both frontier model tokens and cheaper tokens have product market fit. This means the application layer is finding ways to use both and generate ROI with both types Demand for tokens IS demand for compute. This is why SpaceX is looking to build 10GW of compute by next year, because the demand is clearly here Now combine this demand set up, with NVIDIA yesterday announcing financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilize over $500 billion of third party capital for AI infrastructure And Jensen has said publicly he expects $3 to $4 TRILLION of AI infrastructure spend by 2030 The build out will have to continue for a lot longer than the market is expecting, that is very clear to me. Don't let this consolidation period in AI infra stocks shake you out, they will have their moment again and take their next leg higher p.s. if you want to see how im investing in this, you can track my real-time portfolio and the research of all 5 Milk Road PRO analysts with live trade notifications, and it's just $1 to try it out (insane price just to check it out). Learn more here: Good luck out there!show more

The Dawn of a New Era on $SUI (9)... Still in the festive spirit, let’s look at Tusky , Tusky is a storage service that's not controlled by one company. It uses something called WalrusProtocol to keep your data safe. Your data is encrypted from start to finish, so only you can see it. Instead of one place, your data goes to many different spots. This setup makes your data less likely to be lost or stolen. It helps keep your information safe and always available. Tusky gives you control over your files. You can easily manage them with the tools provided. You decide who gets to see your data. This makes it great for personal storage or working with others. It works with SuiNetwork for even more privacy. You can log in without sharing personal info, keeping everything more secure. This means only you can get to your data, with no third party involved. It is growing fast, with 100,000 uploads already. This indicates its increasing acceptance in the tech community focused on data sovereignty. It's good at managing lots of data safely. In tech, where you want to own your data, TuskyTools is popular. It gives users control over their information. This platform helps keep your data secure and gives you freedom.show more

If you think this is just another silly demo... made with AI, read this post. You might change your mind, because this demo is about MATH. What you see on the screen is not a render from Blender (obviously, it’s not that good). It’s a three.js app built with Toolcraft. Available on the web and rendered in real time(link in the comments). But Blender still has a lot to do with it. Blender has Geometry Nodes - a powerful node-based system for creating and manipulating procedural geometry. In other words, it’s math. And math is a universal language. And who do you think is pretty good at math? >>> AI. Now you can download or buy Blender files from marketplaces, and when they contain Geometry Nodes for procedural animations, objects, surfaces, or effects, you can transfer that logic to the web. Make it real-time, make it interactive. Materials are a separate story, of course. They can still suck unless you use the right tricks: PBR, HDRIs, material blending, displacement, and faked surface relief. So why is Blender important here? Blender is open source, and many tools around it are open source too. An AI trained on their code. That means it can translate the math from one environment to another quite accurately. If you’ve been struggling to reproduce some idea with AI that you had in your head or seen in some references, and it has something to do with Geometry Nodes, and you can find that idea or a close one in the Blender ecosystem - it means you can transfer it to the web. Thank me later.show more

Here’s how Starlink next‑generation V3’s laser network works in... space Each Starlink V3 satellite is equipped with six high-capacity space lasers capable of operating at up to 400 Gbps These laser links connect satellites directly to one another, creating a fast, high-bandwidth and resilient mesh network in orbit that can route massive amounts of data around the planet without relying entirely on nearby ground stations That means more network capacity, faster data routing, greater reliability and stronger connectivity across remote and hard-to-reach regions Starlink V3 is not just a more powerful satellite It is a major upgrade to the internet infrastructure being built in spaceshow more

The future of footwear may not be manufactured in... bulk. It may be fabricated around you. That is what makes this shift so interesting to me. 3D-printed footwear is moving from novelty to a real industrial model, with market forecasts pointing to rapid growth over the next decade. At the same time, brands and manufacturers are using additive manufacturing, digital design, and custom-fit workflows to shorten development cycles and make more personalized products viable. What is new here is not just the printer. It is the system around it: → scan the foot → model the fit digitally → print the part on demand → produce closer to the customer That matters. Because once footwear becomes data-driven and locally fabricated, several things change fast: → fit gets more personal → prototyping gets faster → waste drops because you do not overproduce → inventory pressure falls because you do not need to guess demand the same way To me, that is the bigger signal. This is not just about a better sneaker. It is about a different manufacturing logic. Formlabs notes that 3D printing already enables customized orthotics with better biomechanical precision, lower material waste, and simpler digital workflows. McKinsey has also pointed to digitization and 3D design as a way to shorten design cycles and reduce sampling iterations in apparel and footwear. And once that logic matures, the use cases get much bigger: → custom athletic footwear built from gait and pressure data → hospitals producing orthotics faster and closer to the patient → micro-factories making products on demand instead of stocking shelves → footwear designed for one body, not an average body That is why I think this matters now. The question is no longer whether personalized fabrication is possible. It is whether brands move fast enough before customers start expecting every product to fit like it was made only for them. Would you actually wear a shoe fabricated around your own biometric data? #AI #3DPrinting #Footwear #Manufacturing #Innovation #FutureOfWork #RetailTech #Customization #Technologyshow more

Tom Brady is one of the greatest to ever... play the game. His place in football history is locked in forever. On Saturday he posted on Instagram celebrating his son Jack’s 19th birthday — and openly admitted the kid has him beat in an athletic feat he never could have pulled off at that age. That kind of honesty is rarer than championships. Most legends protect their image. Brady just told the truth. Being able to accurately see reality — even when it means your own son has already surpassed you in something — is a higher form of greatness. It takes real confidence to say it out loud and mean it. Brady will be remembered for what he did on the field. Moments like this show why he’ll also be remembered as a father who could look at the truth without flinching.show more

What if you kept asking an LLM to "make... it better"? In some recent work at FAIR, we investigate how we can efficiently use RL to fine-tune LLMs to iteratively self-improve on their previous solutions at inference-time. Training for iterated self-improvement can be costly. The naive approach to training for K self-improvement steps leads to K times the number of rollout steps per episode. We introduce Exploratory Iteration (ExIt), an RL-based automatic curriculum method that bootstraps diverse training distributions of self-improvement tasks by upcycling the LLM's own responses at previous turns as the starting points for both self-improvement and *self-divergence.* In order to decide what task to train on next, the curriculum prioritizes sampling of partial turn histories that led to higher return variance in its GRPO group (a learnability score that comes for free). This automatic curriculum over the bootstrapped task space teaches the model how to perform iterated self-improvement while only ever training the model on single-step self-improvement tasks. We look at ExIt's impact in both single-turn (contest math problems) and multi-turn (BFCLv3 multi-turn tasks), as well as MLE-bench, where the LLM is run in a search scaffold to produce solutions to real Kaggle competitions. Across these eval settings, we find ExIt produces models with greater capacity for inference-time self-improvement compared to GRPO. Notably, ExIt models can self-improve on test tasks for many more steps than the typical solution depth encountered during training, including a 22% improvement in MLE-bench performance compared to GRPO.show more

A few days ago I hinted that big news... is coming. Now, it’s official: I inked a partnership with @pumpdotfun to provide the most valuable trenching resources Solana has ever seen. The video below is a little taste of what’s coming. Thanks to I’ll never charge a dime for it. It’s free, and always will be. On a bittersweet note, that means I’ll no longer be active on Vector. It’s been a blast partnering with the VECTOR crew over the past few months. I’m grateful for the opportunity I had to support them and excited to see what their future holds. If you’ve been missing my broadcasts, don’t worry. I’ll save you a seat at the table. See you in the trenches again soon.show more

Free NVIDIA GPU with 16 GB VRAM GPU for... Running Local LLMs! If you want to master local LLMs but you're waiting until you can afford a $1,500 GPU, you're honestly not going to make it. The open source AI ecosystem is moving way too fast for you to wait on your budget to catch up. Especially when you can build a bleeding edge inference engine from scratch right now, completely for free. You don't need a heavy local rig to start. Google is literally letting you use an enterprise grade NVIDIA Tesla T4 GPU for $0/hour. At standard cloud computing rates (~$0.20/hr), Google Colab’s 4 hour daily free tier hands you roughly $24 worth of data center tier GPU compute every single month. And most people just waste it. Let’s talk about the hardware you get access to for free. The NVIDIA Tesla T4 is an absolute workhorse: - Architecture: NVIDIA Turing (TU104) - VRAM: 16GB GDDR6 (320 GB/s bandwidth) - Compute: 320 Tensor Cores | 2560 CUDA Cores - Performance: 130 TOPS INT8 | 8.1 TFLOPS FP32 - Power: Sipping energy at a max 70W TDP This is the exact same hardware I used to run DeepMind's Gemma 4 26B A4B QAT MoE at a 250,000 context window without a single Out Of Memory (OOM) crash. If you have a web browser and 10 minutes, you have everything you need. I’ve put together a fully documented, cell by cell Google Colab notebook that teaches you exactly how to do this. Here is what the notebook actually teaches you: - How to provision an Ubuntu Linux environment with CUDA 13.0 and verify your driver stack. - How to pull the source code and compile the latest llama.cpp C++ binaries from scratch, specifically optimizing the build for your exact GPU using the -DCMAKE_CUDA_ARCHITECTURES=native flag. - How to directly download quantized local LLMs (GGUF format) straight from HuggingFace using the CLI. - How to manage 16GB VRAM limits, offload neural network layers to the GPU, and push massive context windows. Compile raw llama.cpp, ollama run a model, or spin up the LM Studio CLI. Pick whatever stack you are comfortable with. just start building. No hardware. No credit card. No excuses. Bookmark this post right now so you don't lose the tutorial. Even if you don't have time to run it today, you are going to want this workflow in your engineering toolkit. The link to the free Colab Notebook is in the comments below. Lemme know if you need more tutorials like this.show more

Right now, you may not have access to models... like GPT‑5.6 Sol, GPT‑4.6 Terra, GPT‑5.6 Luna, Claude Mythos 5, or Claude Fable 5. But you can run something surprisingly powerful today, locally, and completely free. in the next 10 mins on your 8 GB VRAM gaming laptop. Gemma 4 26B A4B QAT (MoE) delivers strong performance on a standard 8 GB VRAM GPU using Ollama, with no API, no usage limits, and no external dependencies. Out of the box, it reaches around 20 tokens per second without any optimizations. Only one command in your terminal: Ollama run gemma4:26b This means: Full offline capability (privacy by default) Zero recurring cost Competitive performance for many real world tasks Fast enough for interactive use on cheap consumer hardware If you're waiting for cutting edge cloud models, you're missing what is already practical today: a capable, local LLM that runs entirely on your own machine.show more

Being sore all the time is not a badge.... It means: - Your volume is too high - Your recovery is in deficit - Your next session is compromised - Your nervous system is still cleaning up the last one - Your growth is being throttled by accumulated damage Soreness is a sign you did something. It is not a sign you did the right thing. A trained muscle in a sensible programme should rarely be sore at all. The first few weeks of a new stimulus, yes. After that, your body adapts. The soreness fades. The growth carries on without it. People treat this as a problem to solve. They add drop sets. They add finishers. They chase the burn. They want that next-day ache back because the ache feels like proof. It is not proof. It is damage your body now has to repair before it can build anything new. The lifters making the most progress walk out of the gym feeling worked but not wrecked. They train the same muscle again 72 hours later because they can. The sore-every-day brigade train it again in a week and call it advanced programming. It's just bad arithmetic.show more

He was abandoned when he needed help the most.... Left alone, injured, and struggling to survive, this brave dog lost part of his body — but he never lost his will to live. Today, he is safe in rescue care, learning again what kindness feels like, wagging his tail despite everything he has been through. His journey is not defined by what he lost, but by how fiercely he keeps choosing joy. He still tries to move forward, still seeks affection, still believes in people. Now he needs our help to take the next step in his recovery — a wheelchair that will allow him to move freely, explore the world again, and live without constant strain on his body. A wheelchair means independence. It means dignity. It means this survivor can run, play, and experience life the way every dog deserves. Let’s come together and make America a better home for dogs — a place where abandoned animals are given second chances, where compassion replaces neglect, and where no survivor is left behind. If you can’t donate, please share his story. Your kindness can help change his future.show more

New Version of HyperStore is now live! 🔥 We’re... excited to announce that HyperStore has officially been upgraded to a new system version. This is not a simple UI update, it’s a full platform evolution. ⚡ What’s new? HyperStore now delivers a significantly faster and more intelligent experience powered by its rebuilt infrastructure. - 5000+ AI apps, fully structured into a living ecosystem - A new prompt-based discovery system - Faster navigation, cleaner interface, smarter results Now users don’t search for tools, they instantly reach solutions. 🧠 HyperClaw Integration HyperClaw is now fully active within HyperStore. It acts as a continuous intelligence layer that: - Keeps the platform updated in real time - Curates and optimizes AI apps dynamically - Ensures the ecosystem is always evolving 🔥 What this means? HyperStore is no longer just an AI marketplace. It is now an AI execution layer. Designed for builders, creators, and operators who move fast. 🌐 Try it now: ⚡ Find any AI solution. Instantly.show more

🏝️ Bali's trash burning problem has just reached new... toxic heights This week, Bali's biggest landfill has been ordered to close. The landfill is a more than 35 meter high mountain of decaying trash covering 32 hectares. It's closed because the gases emanated from it are toxic and ground water is being contaminated This means there's no other landfill of this size near to accept Bali's current trash. So the government has instructed people that "household waste should be disposed of at home" which in Bali means burning it! Trash burning has been a tradition for centuries in Bali (and Indonesia), but it used be mostly organic matter that was being burned Once plastic arrived the tradition didn't change though and locals started burning plastic too, pumping toxic gases into the air causing massive spikes of lung cancer and other respiratory diseases The tradition means on a daily basis around 7 in the morning and 6 in the evening, your neighbours will be burning their trash, and if you're not lucky there's a construction site near which will burn even more but all day! Burning plastic means you emit dioxins and furans, some of the most potent human carcinogens that exist I have friends in Indonesia with family with lung cancer cases, it's a real thing. My fear with Bali has always been that the digital nomads there might not realize the slow danger creeping up on them. You get used to the trash burning and polluted air very quickly in Bali and it probably doesn't affect you over a year or so But if you're there for many years, it will in some way or the other! For a community that's so about fitness and health, air quality is weirdly a consideration mostly overlooked in Balishow more

90% of "AI developers" just download pre packaged GGUF... files from Hugging Face, hit run, and call it a day. The top 10% know how to pull the raw safetensors, run the math, and quantize massive models into Q4_K_M themselves. If you think llama.cpp can only execute models, you’re missing the best part of the open source ecosystem. It’s a high performance optimization suite. Manually stripping 69% of the VRAM footprint off a brand new model architecture is where real infrastructure value is made. If you want to actually master local inference and deploy models like Google’s massive Gemma 4 12B it on consumer NVIDIA hardware using llama.cpp, you need to learn this pipeline. Let's build it. I just took the raw 22.7 GB Gemma 4 baseline and manually compressed it down to a 7.02 GB Q4_K_M GGUF artifact using llama.cpp. That is a 69% reduction in footprint. No quality loss. No VRAM bottlenecks. Just native, hardware accelerated C++ inference running a full 2,50,000 token context window on a dual NVIDIA Tesla T4 setup. Stop melting your VRAM on unoptimized weights and stop relying on other people's pipelines. Own your stack. I mapped this entire architecture from dynamic binary fetching to raw quantization and real time GPU streaming into a single, bulletproof notebook. Notebook link is in the comments below. Bookmark this blueprint for your next deployment and tell me which quantization works best for your workflow and model.show more