Posts

Featured Post

My AI lab in a box {or} how I foresee the AI Desktop Future

Image
  Copyright: Sanjay Basu I am running llama.cpp on NVIDIA DGX Spark The NVIDIA DGX Spark just made desktop AI supercomputing accessible. This compact mini PC delivers 1 petaflop of AI performance with 128GB of unified memory . Enough to run models up to 200 billion parameters locally using llama.cpp. It’s bringing data center capabilities to my desk, and the implications are profound for anyone serious about local AI development. Why does this matter? Because for the first time, developers, researchers, and enterprises can fine-tune 70B parameter models and run inference on 200B parameter models entirely on their desks, without data center dependencies, API costs, or data leaving their infrastructure. The DGX Spark paired with llama.cpp’s optimized inference engine creates a sweet spot. Powerful enough for serious work, accessible enough for individuals, and private enough for sensitive applications. While memory bandwidth at 273 GB/s creates some trade-offs compared to discr...

Lessons History Teaches Those Who Listen

Image
  Copyright: Sanjay Basu Mirror Neurons, Mirror Minds — — Week 3 “Those who cannot remember the past are condemned to repeat it.”  — — George Santayana “History is a vast early warning system.”  — — Norman Cousins Tragedy, Farce, and Clickbait History repeats itself, first as tragedy, then as retargeted clickbait. That’s Karl Marx’s old line updated for our age of algorithmic déjà vu. A coup in some fragile democracy looks oddly like another one we scrolled past three years ago. The rise of a demagogue feels suspiciously familiar, like a sequel nobody asked for. Even our memes recycle ancient tropes, as if Plato’s cave has been rebranded as a TikTok filter. But what happens if we actually listen to history? Not skim it, not cherry-pick it, not weaponize it for political soundbites, but hear it. Hear its patterns. Hear its pain. Hear the parts we would rather forget. History, after all, doesn’t shout. It whispers. And in those whispers lie both our warnings and our salvation. The Ec...

Profiling: An Evolutionary Shortcut or a Moral Dead End?

Image
  Copyright: Sanjay Basu Mirror Neurons, Mirror Minds — Week 2 “The measure of a man is what he does with power.” Plato “Prejudice is a burden that confuses the past, threatens the future, and renders the present inaccessible.” Maya Angelou A Thought You Don’t Want to Admit We all profile. That’s not an accusation. It’s biology. It’s survival. Your brain does it before you even know you’ve had a thought. Someone steps into the elevator and you scan height, clothing, body language, tone of voice, all in a fraction of a second. It’s not malice. It’s pattern recognition. A vestige of the days when mistaking a rustling bush for wind instead of a predator meant you didn’t live to pass on your DNA. The speed at which this happens is staggering. Research from Princeton psychologists Janine Willis and Alexander Todorov found that we form judgments about trustworthiness in as little as 100 milliseconds, faster than a blink. These split-second assessments influence everything from who we sit...

ChatGPT Just Pulled the Rug Out from under n8n, Make, and Zapier

Image
  Copyright: Sanjay Basu Here’s Why AgentKit Kills the Middleman A single update just vaporized an entire swath of startups. Zapier, Make, n8n. All the automation glue tools you once trusted, got a black swan-sized threat when OpenAI dropped AgentKit. Overnight, the promise of drag-and-drop agent creation turned from a clever thought exercise into a working, production-grade system. The era of stitching APIs by hand feels suddenly archaic. Yes, I’m claiming that ChatGPT just made startups obsolete. Or, at least, drastically reshaped which ones survive. My opinion, of course! The timing is uncanny. We’ve long lived in an era of “no-code” and “low-code” optimism. Anyone can build your integration, pipeline, or workflow without touching lines of code. Tools like Zapier, Integromat (now Make), and n8n democratized automating tasks. They bridged silos, mapped data flows, and let you chain triggers and actions across apps. But they were never built for the era of autonomous, agentic AI. ...