Sunny

← All weeks

What I’ve been reading (W50, 2025)

LLMs as simulators, programming in the AI era, scaling Google Kubernetes Engine and fun lists

  1. LLMs are simulators, not thinkers. Instead of asking “What do you think about X?”, ask “What kind of people would discuss X, and what might they say?” LLMs don’t have ongoing thoughts or opinions. When you say “you,” they just pick a personality based on patterns in their training data and respond in that style https://x.com/karpathy/status/1997731268969304070

  2. Stay curious, think in systems, communicate with precision, own the quality of what you ship, and approach things with a polymath mindset to thrive as a developer in the AI‑assisted era https://www.youtube.com/watch?v=3Y1G9najGiI

  3. Google ran a 130,000-node GKE (Google Kubernetes Engine) cluster experimentally, and at this scale, it sustained 1,000 pod operations per second and 10s P99 startup times. This shows how much Kubernetes can scale to mega clusters while keeping latency and stability under control https://cloud.google.com/blog/products/containers-kubernetes/how-we-built-a-130000-node-gke-cluster

  4. Nano Banana can be prompt-engineered for incredibly nuanced AI image generation https://minimaxir.com/2025/11/nano-banana-prompts

  5. A fun website to see how the size of life scales (turn on the sound for a better experience!) https://neal.fun/size-of-life