LLM Inference Handbook

A practical handbook for engineers building, optimizing, scaling and operating LLM inference systems in production.

Repasting a MacBook

Postgres LISTEN/NOTIFY does not scale

Postgres LISTEN/NOTIFY can cause severe performance issues under high write concurrency due to a global lock during commit. Learn why it doesn't scale and how to avoid outages.

Show HN: Interactive pinout for the Raspberry Pi Pico 2

An interactive, accessible and beautiful GPIO Pinout guide for the Raspberry Pi Pico 2

OpenFront: Realtime Risk-like multiplayer game in the browser

FP8 is ~100 tflops faster when the kernel name has "cutlass" in it

Kimi K2 is a state-of-the-art mixture-of-experts (MoE) language model

Upgrading an M4 Pro Mac mini's storage for half the price

I'm done with social media – Or: why I have a blog now

Or: why I have a blog now.

Conspiracy theorists unaware their beliefs are on the fringe

Overconfidence is a hallmark trait of people who believe in conspiracies, and they also significantly overestimate how much others agree with them, Cornell psychology researchers have found. 

Overtourism in Japan, and how it hurts small businesses

Why being popular is not the goal of most small businesses

Batch Mode in the Gemini API: Process More for Less

Discover how Batch Mode in Gemini API allows developers to submit large jobs, offload processing, and get results in 24 hours, offering simplicity, convenience, and cost-savings.

Recovering from AI addiction

Things I learned from 5 years at Vercel

Here are some of the lessons I learned and how I grew as a leader and manager, as well as a bunch of behind-the-scenes photos.

At Least 13 People Died by Suicide Amid U.K. Post Office Scandal, Report Says

A public inquiry into the wrongful prosecutions of about 1,000 postal workers has uncovered more victims than previously known, according to a report.

Bill Atkinson's psychedelic user interface

How an Apple legend approached the world’s most powerful psychedelic as a user-experience problem to be solved.

AI agent benchmarks are broken

Benchmarks are foundational to evaluating the strengths and limitations of AI systems, guiding both research and industry development.

Show HN: Vibe Kanban – Kanban board to manage your AI coding agents

Kanban board to manage your AI coding agents. Contribute to BloopAI/vibe-kanban development by creating an account on GitHub.

Top DNS domains seen on the Quad9 recursive resolver array each day

Top 500 DNS domains seen on the Quad9 recursive resolver array each day - Quad9DNS/quad9-domains-top500

Show HN: Pangolin – Open source alternative to Cloudflare Tunnels

Tunneled Reverse Proxy Server with Identity and Access Control and Dashboard UI - fosrl/pangolin