IInference Engineering Academy10h ago · 4 min readPrefill vs decode: why LLM inference is memory-bandwidth boundAsk why an LLM streams its answer at a steady ~100 tokens/s even on a GPU rated for nearly a petaFLOP, and you end up at one idea: generation has two phases with very different bottlenecks. Two phases1inferenceacademy.hashnode.dev
UUrban Mixo6h ago · 7 min readThe 14KB Design System: Styling 30 Interactive Web Apps with Zero Frameworks and 100/100 Core Web VitalsOpen the source code of almost any modern web application, and you will find a familiar styling stack: Tailwind CSS with thousands of generated utility classes, a CSS-in-JS runtime like styled-componeurbanmixo.hashnode.dev
JJulian Neagu8h ago · 12 min readI Rebuilt My Portfolio as a Cinematic 3D Website You Fly ThroughKey Highlights: A cinematic 3D website is a real 3D world navigated through smooth camera movement, like flying through a short film. Built with WebGL, it's the most memorable web experience you can cjulianneagu.hashnode.dev
RRavi Sah14h ago · 6 min readHow to Stop Deadlocks in SQL Server: 5 Production Patterns and Their FixesHow to Stop Deadlocks in SQL Server: 5 Production Patterns and Their Fixes We’ve all been there: a production application suddenly stumbles, errors spike, and your telemetry logs reveal the dreaded Erprodsolutions.hashnode.dev
AAlki15h ago · 4 min readiPhone Safari crashes or reloads on my scroll animation: why and how to fix itIf your scroll image sequence makes iPhone Safari reload the page or show 'A problem repeatedly occurred', the tab is running out of memory. Here's why decoded frames cause it and how to fix it. Shortscrollsequence.hashnode.dev
RRoshan Chaturvedi19h ago · 19 min readHow React Native Apps Achieve Near Native PerformanceThink of a phone screen as a stage. Every sixteen milliseconds the curtain rises on a new frame, and the audience, which is the person holding the device, judges the show without ever saying a word. Wjustacademyofficial.hashnode.dev
DDip Roy1d ago · 3 min readFrom the 6:07 AM Local Train to the Developer Terminal: My Daily Passenger JourneyEvery morning before dawn breaks over Krishnanagar, the world is quiet. But for a daily passenger heading out toward Kolkata, that calm never lasts long. Standing on the station platform with sleep stdiproy.hashnode.dev
SSudhanshu Wani1d ago · 13 min readSensor to screen in under 100 ms: building and breaking a Kafka pipelinePicture a control-room display for about 20 live machines. When one of them changes, the screen should show it before you can blink. I built a system that does this, measured how fast it really is, thtechblog01.hashnode.dev
DDan Holloran1d ago · 5 min readGraphQL @defer: Stop Letting Your Slowest Field Set Your Load TimeThe pitch for GraphQL is that a page asks for everything it needs in one request. The catch nobody puts on the slide: that one request is only as fast as its slowest field. Your product page has name,grimicorn.hashnode.dev
SSachin Rajput1d ago · 10 min readThe KV Cache From ScratchIn every part of this series, one thing kept showing up: the notebook. In Part 1 it grew by 72 KB with every word. In Part 3 it was the part of the work that batching couldn't share. In Part 4 it staysachinrajput.hashnode.dev
