Bonsai 27B on RTX PRO 6000 vs DGX Spark: what actually works(opens in new tab)
Real Bonsai 27B benchmarks on an RTX PRO 6000 and a DGX Spark, including the supported llama.cpp setup, ternary vs 1-bit results, and speculative decoding.
Curated river of news
Reader
Latest posts from blogs I follow
2985
entries
60
pages
Workshop Basel day two(opens in new tab)
If you missed it. I already described day one. Caffeinated and ready, we all gathered in the same spacious room as yesterday, but seated in new places as “suggested” by our captain. Some of us even remembered to move over ...
How a Junior Developer Created One of Java's Most Widely Used Server Technologies(opens in new tab)
Understanding the Rust hype for the busy developer(opens in new tab)
First, let's be clear, this post is not intended to fuel any programming language holy war. I wrote this post because I've sometime been late to understand the benefits of
Web-Perf Wednesday 001 – SPAs Are Finally Becoming Measurable(opens in new tab)
Chrome 151 brings native soft-navigation measurement closer, alongside better redirect timing, revealing CrUX data, and more trustworthy browser tooling.
How Much I Spend on AI Agents as a Software Engineer (And Is It Worth It?)(opens in new tab)
I've been leaning into agentic engineering for the past six months, and I've finally hit my stride at work after coming back from parental leave. AI token spend has been a hot topic and we have intern...
reCAP 2026 talk resources(opens in new tab)
usbtree(opens in new tab)
A live USB device tree TUI with Linux activity metrics.
Dell is on a roll with the XPS(opens in new tab)
Workshop Basel day one(opens in new tab)
On this hot summer’s day in Basel, Switzerland, the seventh HTTP workshop started. These events tend to work roughly the same way and the people in the room are also to large extent familiar and known since previous editions. Forty ...
The Three Components of Becoming AI Antifragile(opens in new tab)
The three components of becoming AI antifragile/images/becoming-ai-antifragile-header.webp/images/becoming-ai-antifragile-header.webp I have a new idea that everything you should try to do to get ready for AI basically breaks down into three pieces. Deeply understanding how things work The firs...
ISO 9001 and AS9100: What They Actually Buy You(opens in new tab)
If you've spent any time around a machine shop or a manufacturing floor, you've probably heard someone grumble that a quality certification is just a binder full of paperwork nobody reads until an auditor shows up. That reputation isn't entirely ...
I made a Plex server(opens in new tab)
There’s a lot to like about Plex from a philosophical standpoint — it’s local-first software centered around sharing your favorite art with your friends. Plex hosts movies, TV, and music — media you purchase because you love it, and share ...
Building a Custom Metrics Exporter for Kubernetes(opens in new tab)
Kubernetes ships with built-in awareness of CPU and memory, but most real-world scaling decisions depend on signals that live entirely outside that narrow window: how many messages are waiting in a queue, how long the last batch job took, how ...
France Deleted Windows, Now They Use THIS Unpopular OS(opens in new tab)
#488: tau - it's 2pi and it writes code(opens in new tab)
Topics covered in this episode: The trusted-publishing debate: how to do it right vs. why you shouldn't trust it JupyterLab 4.6 and Notebook 7.6 are out! Tau – new small, readable terminal coding agent Django Tasks and Django 6.1 Extras ...
The discourse about software craftsmanship conveniently ignores distribution(opens in new tab)
I've always considered myself a software craftsman. I love clean APIs, efficient code and great user experience. If our entire world is about to be rewritten in Rust governed by
I have no idea who celebrities are anymore (opens in new tab)
Julia Roberts? She was in that one movie with that guy, and the other one with the other guy, and like 100 more. Whatever. But she’s old news. Like all...
Logo: Squawkbox(opens in new tab)
Operating AI/ML Workloads on Kubernetes: A Headlamp Plugin for Kubeflow(opens in new tab)
Kubernetes has quietly become the default platform for AI and machine learning. Whether you run notebook servers for data scientists, schedule distributed training jobs, tune hyperparameters, or orchestrate multi-step ML pipelines, those workloads increasingly land on a Kubernetes cluster. Kubeflow ...
Kubernetes Dashboard to Headlamp: A Step-by-Step Guide(opens in new tab)
1. Before you start: know what is changingKubernetes Dashboard and Headlamp both show what is running in a cluster, but they work differently. When Headlamp runs on the desktop, it uses your existing kubeconfig to connect to one or more ...
Marimo Pair - A Canvas for Agent + Developers Collaboration(opens in new tab)
Coding agents have gotten really good at one kind of work. You scope a feature, edit some files, run the tests, ship it. It all happens on disk. But that is not how data work feels. You load something, you ...
Avoid the AI Expertise Trap(opens in new tab)
Avoid the AI expertise trap header/images/avoid-the-ai-expertise-trap.webp/images/avoid-the-ai-expertise-trap.webp The further away a topic is from your expertise, the smarter an AI will sound. This is a blind spot that not enough people are aware of or take seriously. The best possible way ...
itertools cheatsheet(opens in new tab)
Cheatsheet with visual diagrams that explain how the iterables from itertools work. This cheatsheet contains diagrams that explain how the iterables from the module itertools work in a visual way. Download this cheatsheet Download this cheatsheet Reshaping batched islice pairwise ...
Laughing Through The Pain with David Crawshaw(opens in new tab)
Some of you might know David as a co-founder of Tailscale. In this episode we're talking about a new kind of cloud he's building at exe.dev. The SSH-only interface is a breath of fresh air compared to the monstrocity that ...
Even if 90% of AI-discovered vulnerabilities are made up, it's still terrifying(opens in new tab)
Developers working on popular projets are inundated with hallucinated vulnerability reports, leading people to downplay the risks associated with AI-assisted vulnerability hunting. But hear me out: it should wake you
The Tower Keeps Rising(opens in new tab)
I feel that some vibecoded software changes somewhat randomly and unexpectedly. That made me think about Bruegel’s “The Tower of Babel” which shows an already quite chaotic depiction of the Tower of Babel. The story is one of pride and ...
But Y(opens in new tab)
From Prompt Engineering to Intent Engineering(opens in new tab)
Intent Engineering/images/intent-engineering.webp/images/intent-engineering.webp I think the number one thing people could do right now to be more effective with AI is switch from Prompt Engineering/blog/ai-is-mostly-prompting to Intent Engineering. This is due to Sutton's Bitter Lessonhttp://www...
Minimizing Machine Work Maximizes Human Work(opens in new tab)
Minimizing machine work maximizes human work header/images/minimizing-machine-work-maximizes-human-work.webp/images/minimizing-machine-work-maximizes-human-work.webp Behavioral geneticists like Robert Plominhttps://en.wikipedia.org/wiki/Robert_Plomin have shown that if you minimize environmental d...
On Atproto and Atmospheric Groups(opens in new tab)
I'm starting A Fresh Blog(tm), this time on Pckt. I wanted a dedicated space to post about development, atproto, and community thoughts, and a new space felt like the right place for it. Plus I wanted to get in on ...
AI Blogging From Inside Vim(opens in new tab)
A human hand writing a manuscript while a mechanical hand places a small patch of text with tweezers/images/ai-blogging-from-inside-vim.webp/images/ai-blogging-from-inside-vim.webp Daniel didn't write this one. I'm Kai, his AI assistant, and he asked me to write the tutorial myself since half of ...
Recreating the Bell Labs Cafeteria(opens in new tab)
Bell Labs cafeteria header/images/recreating-the-bell-labs-cafeteria.webp/images/recreating-the-bell-labs-cafeteria.webp Anthropichttps://www.anthropic.com's Claude is named after Claude Shannonhttps://en.wikipedia.org/wiki/Claude_Shannon. I've read a couple of biographieshttps://en.wikipedia.org...
Self-Host Weekly (10 July 2026) (opens in new tab)
Everybody's mad, nobody's leaving
Momentum is why the spice continues to flow(opens in new tab)
People are sometime wondering how I can be so "productive" with my blog (granted that publishing brainfarts can be called "produtivity"). Come here, I will tell you my secret. I
Trustworthy AI in Healthcare and Longevity(opens in new tab)
You ask an AI a question and it answers with total confidence. Most of the time, a confidently wrong answer is just an annoyance. But what if the question is medical, and there's a real patient on the other end? ...
I Think AGI Just Happened(opens in new tab)
Claude Tag AGI moment header/images/claude-tag-is-agi.webp/images/claude-tag-is-agi.webp I think we just saw the birth of AGI, and it's from the most unexpected place. At least for me. I think it arrived in the form of a product from Anthropichttps://www.anthropic.com called Claude Taghttps://cla...
smallvec is probably one of the most underrated Rust crates(opens in new tab)
Hear me out: While Rust, the language, gives you all you need to build libraries and programs that can scale from microcontrollers to big servers and everything in between, in
Announcing etcd v3.7.0(opens in new tab)
This article is a mirror of the original announcement Today, SIG etcd is releasing etcd v3.7.0, the latest minor release of the popular distributed key-value store and core Kubernetes component. v3.7 ships the long-requested RangeStream feature, delivers several other performance ...
How to Use AI Coding Tools Well(opens in new tab)
How to use AI for coding well? Practical tips for using AI coding tools, including context management, AGENTS.md files, planning, reviews, tests, and MCPs.
Setting up Sveltia CMS for an Astro Markdown Blog(opens in new tab)
How I added Sveltia CMS to my Astro markdown blog. Setup, image optimization, mobile editing fixes, and solving frontmatter issues.
The Grammar of Data: Define Once, Run Anywhere with Cross-Engine Expressions(opens in new tab)
Grammars for languages or any other field are a beautiful thing. They compress complex systems into a language with a couple of rules. For the spoken language example, we know when to capitalize a letter or how to start a ...
Investigating why RustCrypto is slow: Deep dive into SIMD instructions and hardware acceleration(opens in new tab)
Now that I have your attention, let's be clear, RustCrypto packages are not "slow", they are most probably "fast enough" for your use case. Yet, I couldn't unsee how faster
2026.H1 Review(opens in new tab)
TL;DR - In 2026.H1, I started feeling like a real dad, took parental leave, went to a programming retreat, shipped CloudSeed Rust + QuietCut, reached 130k views, got back to building for fun, and star...
Computing memories(opens in new tab)
dnsglobe(opens in new tab)
A global DNS propagation checker for your terminal.
How to publish to PyPI using GitHub Actions securely(opens in new tab)
There have been several security incidents lately that involved compromising GitHub Actions workflows. This has led some to say "GitHub Actions is the weakest link" in publishing and to GitHub publishing a GitHub Actions security roadmap update. But saying it's ...
The Real Reason You Can't Get a Tech Job (It's NOT Your Resume)(opens in new tab)
#487: Minimum requirements(opens in new tab)
Topics covered in this episode: dust - a better du Hermes Agent: The AI agent that grows with you llm-coding-agent 0.1a0 Extras Joke Watch on YouTube About the show Sponsored by us! Support our work through: Our courses at Talk ...
