EZThrottle — API Rate Limit Router
A Queue Per User at Scale
Announcing AccountQueues: per-user, per-key queue isolation, provider flow control headers, and why 2 RPS default protects the internet.
Serverless 2.0: RIP Operations
What happens when you stop managing retries and start shipping features. Scale requests, not pods.
EZThrottle: Making Failure Boring Again
EZThrottle is a multi-region HTTP proxy for external APIs. Race requests across regions, survive partial outages automatically.
Server-Driven UI in 22 lines of TypeScript
Move the layout decision out of the clients and into the API. One JSON contract; every client renders it in its own programming language or framework.
The Boring Internet
The internet you grew up on isn't dying. A commercial veneer glued on top of it is.
Caching, productized
Build caching is becoming an everyone problem. Here is how we think about productizing it at Tuist for indie developers and small teams, not just enterprises.
Working Backwards PR/FAQ Instructions & Template - Working Backwards
Working Backwards PR/FAQ Instructions & Template Working Backwards PR/FAQ Instructions & Template Start with the customer and work backwards—harder than it sounds but a clear path to innovating and delighting customers—a useful working backwards tool: writing the press release and FAQ before you build the product. Introduction Most of Amazon’s major products and initiatives since […]
AI likes deep modules
You should too
How The Heck Does GPS Work? (An Interactive Exploration)
An interactive exploration of how GPS works — from the speed of light and trilateration to the cheap clock problem and Einstein's relativity corrections.
Branch‑Avoidant Programming
Equal Ping for All
No, Really, Bash Is Not Enough: Why Large-Scale CI Needs an Orchestrator - Ian Duncan
A deeper dive into why shell scripts cannot replace a proper CI control plane, drawn from years of building and operating CI systems at scale.
Honker
Durable queues, streams, pub/sub, and time-trigger scheduling on SQLite. One file, zero servers.
How Container Filesystem Works: Building a Docker-like Container From Scratch | iximiuz Labs
Learn how Linux containers are built from the ground up. Starting with the mount namespace and a root filesystem, see why PID, cgroup, UTS, and network namespaces naturally follow - and how this foundation makes concepts like bind mounts, volumes, and persistence in Docker or Kubernetes much easier to grasp.
Agent Memory Patterns
A short HOW TO guide for agent memory systems. Especially the difference between blocks, files and skills.
What the Bootloader Knows — Vivian Voss
FreeBSD loader (Forth, ZFS-aware Boot Environments), LILO (block lists, 1992-2015), GRUB 2 (filesystem drivers for everything). Three loaders, three theories of the operating system.
Trusted vs. Untrusted Agency Amplifiers | Jason Rodriguez
Jason Rodriguez is an interesting mix of marketer, designer, developer, and operations pro based in Michigan. Working at GitHub.
Headless everything for personal AI
Posted on Saturday 18 Apr 2026. 1,331 words, 11 links. By Matt Webb.
Rail-Optimized Networking for AI Training Workloads
Over the past decade, Clos-based leaf-spine architectures have become the default design for data center networking. They deliver predictable latency, horizontal scalability, and clean integration …
Hmmm: Rail-Optimized Networking for AI Workloads « ipSpace.net blog
Phil Gervasi wrote an interesting article describing Rail-Optimized Networking for AI Training Workloads. Go read it first; I’ll wait.
Does it sound interesting? Were you able to see behind the curtain and figure out what it’s really about?
The Dependency Avalanche — Vivian Voss
Express 65 packages. Next.js 644. XZ 2.5 years of social engineering, caught by 500 ms of SSH latency. The audit duty has been outsourced so many times that nobody is holding it.
Shared Nothing Engineering
AI made code free. Coordination is the bottleneck now. The distributed-systems toolkit — partition, shard, shared-nothing — turns out to describe the next decade of org design.
Building a Conversational Interface for Manifold with MCP and Picos
GUIs are dead—at least for most user experiences. This post describes a BYU capstone project where five seniors built a conversational interface for Manifold using MCP and picos. The result shows how natural language can replace a GUI entirely, letting users create, tag, and manage digital things through dialogue instead of learning a standard graphical user interface.
Home - developer - Confluence
The invisible engineering behind Lambda’s network
The subtle inventiveness that reduced cold start setup from seconds to 200μs.
The perfect file upload pattern
Adam Silver – interaction designer – London, UK
FastCGI Specification
This specification has narrow goal is to specify, from an application perspective, the interface between a FastCGI application and a Web server that supports FastCGI.
Source code sandboxing
FastCGI: 30 Years Old and Still the Better Protocol for Reverse Proxies
For FastCGI's 30th birthday, let's look at how it avoids the security problems inherent in HTTP reverse proxying