Blog
Writing on GPUs, LLMs, MLOps, Kubernetes — and mindset · 3517 posts
#2026-03 765#english 592#culture 264#deep-dive 254#kubernetes 247#career 229#ai 216#llm 208#devops 193#2026-04 146#security 141#database 114#observability 113#communication 109#history 107#architecture 100#productivity 96#finance 88#economy 84#mindset 81#psychology 80#ai-papers 79#food 78#it 78#travel 78#deep-learning 77#japanese 77#networking 77#performance 72#business-travel 70#linux 70#gpu 69#ai-agent 66#cs-fundamentals 63#postgresql 60#rag 58#self-improvement 55#learning 53#mlops 53#ai-platform 51
3.3V vs 5V and Level Shifting: Why Raspberry Pi GPIO Can't Survive 5V
Wire a 5V sensor straight into a Raspberry Pi GPIO pin and the pin dies. This post starts from the definition of logic levels to explain exactly why. It covers what the VIH and VIL thresholds mean, why a 3.3V output is u
2026-07-31 · 20 min read #electronics#raspberry-pi#arduino#hardware#iotLED Resistor Calculations Done Right: Why Every Color Needs a Different Value, and the E24 Series
Starting from a diode's exponential current curve, this explains why an LED must always have a series resistor. It tables forward voltage by color, then works out real resistor values for both 5V and 3.3V supplies. It co
2026-07-31 · 21 min read #electronics#arduino#circuit#hardware#raspberry-piSeeing TCP Congestion Control — CUBIC, BBRv3, and a gVisor Netstack Running in Your Browser
On July 28, 2026, "Simulating TCP loss and congestion in browser using Go/WASM" went up on Hacker News. It builds two entire copies of the gVisor netstack to WebAssembly, runs them in browser workers, and puts a bottlene
2026-07-31 · 16 min read #network#tcp#congestion-control#bbr#wasmWhen Open Source Governance Breaks Down — The Ruby Central Dispute and Three Axes of Control
On July 30, 2026, André Arko published "Ruby Central's Destructive Legacy," bringing the RubyGems repository access dispute that began in September 2025 back to the surface after ten months. This post lays out the primar
2026-07-31 · 15 min read #open-source#governance#rubygems#supply-chain#licensingTerminal UI Development Guide — Ink, OpenTUI, Bubble Tea, ratatui, Textual: What to Use When
When termcn — a shadcn-style collection of terminal components — hit GeekNews, TUI development became a hot topic again. Instead of listing frameworks, this post organizes the ecosystem around three rendering models — th
2026-07-31 · 15 min read #tui#terminal#ink#opentui#cliThe Conditions Under Which a 9B Fine-Tuned for 500 Dollars Beat the Frontier — And How Narrow They Are
On July 28, 2026, a post scored 336 points on Hacker News. It reports that Fermisense trained Qwen3.5-9B with GRPO on roughly 500 dollars worth of GPU time and beat five frontier configurations — using the same tools and
2026-07-31 · 11 min read #ai#llm#fine-tuning#reinforcement-learning#inference-costHow a 26B Model Runs in 2GB of RAM — Resident Memory and Working Set Are Not the Same Number
TurboFieldfare, posted to Show HN on July 29, 2026, claims to run Gemma 4 26B-A4B on M-series Macs in about 2GB of RAM. 14.3GB is installed on disk, the only thing resident in RAM is a 1.35GB shared core, and the expert
2026-07-31 · 12 min read #ai#llm#quantization#apple-silicon#moeGemini Robotics 2 and the Robot Foundation Model — What Whole-Body Control Actually Changes
On July 30, 2026, Google DeepMind unveiled Gemini Robotics 2, announcing that a single vision-language-action model now controls a humanoid from its toes to its fingertips. Interestingly, the numbers released alongside i
2026-07-31 · 14 min read #ai#robotics#vla#foundation-models#evaluationSimulating Systems in the Browser — What WASM Makes Possible and What Still Blocks It
ccsim, posted to Hacker News in late July 2026, builds two full copies of gVisor's userspace TCP stack into WebAssembly and runs them in a browser worker. v86 JIT-compiles an x86 CPU to WASM to boot Windows 98, and WebVM
2026-07-31 · 15 min read #webassembly#wasm#simulation#browser#systemsThe Mean Tells You Nothing — How to Debug Latency Through Distributions
Farid Zakaria's post "The mean means nothing," published July 27, 2026 and reaching the top of Hacker News, covers a case where average latency got 9% worse after a caching layer rollout — from 112ms to 122ms. In the ver
2026-07-31 · 14 min read #observability#latency#performance#histogram#percentileChoosing a Self-Hosted Deployment Platform in 2026 — The Five Axes That Separate Coolify, Dokploy, CapRover, Kamal, and Openship
Openship, a self-hosted deployment platform with built-in CI/CD, hit GeekNews and brought back the same old question: leave managed PaaS and run it yourself? This post compares Coolify, Dokploy, CapRover, Kamal 2, Opensh
2026-07-31 · 12 min read #self-hosting#deployment#paas#kamal#coolifyThe Illusion of Talent — What the Practice Research Actually Says, and What Actually Compounds
In late July 2026, "The Illusion of Talent" climbed to the top of GeekNews. The title makes it sound like a scientific dispute about talent and practice, but the original is a personal essay about a junior developer's ca
2026-07-31 · 14 min read #career#deliberate-practice#expertise#psych-papers#mindsetMarkdown/MDX Pipeline Performance — What satteri and Native Cores Actually Change
When satteri, a high-performance Markdown/MDX processor for the JavaScript ecosystem, hit GeekNews, documentation site build times became a hot topic again. This post breaks down where the time in a thousands-of-pages MD
2026-07-31 · 14 min read #mdx#markdown#build-performance#unified#static-siteGPT-5.6 and the Limits of Price-Performance — How to Find Your Workload's Place on the Curve
On July 30, 2026, OpenAI cut GPT-5.6 Luna by 80% and Terra by 20%. That is three weeks after the July 9 general availability, and the top-end Sol price is unchanged. Luna went from 1 dollar input and 6 dollars output per
2026-07-31 · 12 min read #ai#llm#openai#inference-cost#prompt-cachingThe Complete Guide to GitHub Stacked PRs — What Native Stacking Solved, and What It Left Standing
On July 30, 2026, GitHub shipped Stacked Pull Requests into public preview. The workflow of splitting a large change into several small PRs with dependency order, reviewing them in parallel, and merging them together now
2026-07-31 · 13 min read #github#stacked-pr#code-review#git#developer-workflowA Paper With Fake Authors Got Accepted for an Oral — a Systems Analysis of Peer Review's Failure in the AI Era
On July 30, 2026, two reviewers disclosed that of the 22 ML conference submissions they reviewed this summer, 15 contained fabricated citations, nonexistent authors, or unmistakable signs of LLM generation. Two of them c
2026-07-31 · 13 min read #ai#peer-review#research-integrity#llm#academiaThe Truth About Becoming a Manager — It Is Not a Senior Engineer With Meetings Added
"The truth about being a manager," posted to GeekNews in late July 2026, is a personal-experience essay that lists fourteen items. This article explains, structurally, why those items keep recurring. The essence of the m
2026-07-31 · 14 min read #career#engineering-management#leadership#ic-to-manager#feedback-loopWhat Transfers Through Distillation and What Does Not — Distilling DeepSeek into GPT-OSS Did Not Bring the Censorship Along
CTGT's experiment, posted to Show HN on July 30, 2026, reports that distilling financial reasoning from DeepSeek V4 Flash as the teacher into GPT-OSS as the student carried the capability over but not the political censo
2026-07-31 · 12 min read #ai#llm#distillation#alignment#open-weightsMonorepo CI Cache Strategies — Fix the Cache Key, Not the Cache Size
Most of the time, monorepo CI is slow not because the cache is too small but because the cache key is wrong. This post covers what the Turborepo, Nx, Bazel, and GitHub Actions caches actually hash, three levers for narro
2026-07-31 · 14 min read #monorepo#ci-cd#build-cache#turborepo#bazelMitchell Hashimoto's Second Company, Superlogical — What's Public and What Isn't
On July 29, 2026, Mitchell Hashimoto unveiled a new company, Superlogical, and it picked up 776 points and 452 comments on Hacker News. The first product is a terminal multiplexer, built on libghostty as an MIT-licensed
2026-07-31 · 13 min read #developer-tools#startup#ghostty#terminal#infrastructure