System design is changing. A year ago, I would have said system design was about databases, caching, load balancers, and scaling. It still is. But today, more and more systems also call an LLM. Over the past year, I have built several backend projects that power AI features, and I learned fast that AI introduces a whole new set of system design questions: → How do you handle slow model responses? → How do you stream outputs to users? → How do you keep costs under control as you scale? → What happens when the model gets it wrong? If you're a backend engineer, learning AI system design is one of the best moves you can make right now. Here's where I would start: - Grokking the Generative AI System Design: educative.io/courses/generat… - Agentic System Design: educative.io/courses/agentic… And if you want the basics first, I have a video that walks through every important system design concept by scaling one app from 1 to 10 million users: 📺 youtu.be/ikDAN9aDIZg 🍂 Educative's Fall Sale is on: 55% off courses on AI, system design, and more. You code right in the browser, with no setup. #systemdesign #backend #ai #llm #softwareengineering #rust

I can finally share what I have been building. It’s called @ObsidianLog, an open-source log archival tool I built with Rust for structured operational logs. It gives you a separate path for logs you want to retain long term, with local encryption, compression, integrity verification, and the option to archive locally or across Sia’s decentralized storage infrastructure. ObsidianLog is officially in public beta! 🚀