joshuago’s Bookmarks

A set of advanced theoretically grounded quantization algorithms that enable massive compression for large language models and vector search engines.

Built on a fully open, end-to-end data pipeline that spans pretraining, post-training, and interactive reinforcement learning, which gives developers reproducible building blocks.

A data format similar to JSON, but richer and more extensible. Think of it as "JSON with superpowers."

HipKittens is an opinionated collection of programming primitives to help developers realize the hardware's capabilities. Includes optimized register tiles, 8-wave and 4-wave kernel patterns instead of wave-specialization to schedule work within processors. Also includes chiplet-optimized cache reuse patterns to schedule work across processors.

Nice writeup from Atlassian, makers of Jira and Confluence. Includes details about tuning specific JVM parameters, looking for CPU L3 cache misses, and measuring TLB tablewalks.

A modern alternative to htop, with a gamified interface.

Memory-safe implementation of C and C++. Lots of software compiles and runs with Fil-C with zero or minimal changes. All memory safety errors are caught as Fil-C panics.

Overview of DuckDuckGo's hiring practices and what to expect as a candidate.

How do you align a group of people around a shared goal? You make sure the right things are happening. Together. Over and over again.