Nine Things I Learned in Ninety Years
edwardpackard.com
Zoxide: A Better CD Command
github.com
Go has added Valgrind support
go-review.googlesource.com
Qwen3-Omni: Native Omni AI model for text, image and video
github.com
Delete FROM users WHERE location = 'Iran';
gist.github.com
Themis (European Reusable Rocket) is assembled on launch pad
phys.org
I built a dual RTX 3090 rig for local AI in 2025 (and lessons learned)
llamabuilds.ai
Gamebooks and graph theory (2019)
notes.atomutek.org
The YAML Document from Hell
ruudvanasseldonk.com
Processing Strings 109x Faster Than Nvidia on H100
ashvardanian.com
Telli (YC F24) is hiring ambitious engineers [Berlin, on-site]
hi.telli.com
Cap'n Web: a new RPC system for browsers and web servers
blog.cloudflare.com
Compiling a Functional Language to LLVM (2023)
danieljharvey.github.io
Indoor surfaces act as sponges for harmful chemicals
news.uci.edu
I'm spoiled by Apple Silicon but still love Framework
simonhartcher.com
Paper2Agent: Stanford Reimagining Research Papers as Interactive AI Agents
arxiv.org
Why haven't local-first apps become popular?
marcobambini.substack.com
The Beginner's Textbook for Fully Homomorphic Encryption
arxiv.org
The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text
arxiv.org
Is a movie prop the ultimate laptop bag?
blog.jgc.org
> we manually curated a set of over 2,000 YouTube channels that release original openly licensed content containing speech. From these channels, we retrieved and transcribed (using Whisper) over 1.1 million openly licensed videos comprising more than 470,000 hours of content.
This is why Gemini has such an advantage.
Also, link to explore data: https://huggingface.co/collections/common-pile/common-pile-v...