In this post I want to discuss how we organize systems by contrasting two families of approaches. The first is about analytical decomposition that aims to maintain control over a system, and the other is based on a perspective of complex systems that resist analysis, which tend to focus on figuring out interactions and mechanisms to foster desirable emergent behaviour.

Pilots have a specific term for moving from one airplane to another: a "type conversion". Not "learning to fly" again — you already know how to fly. It's the process of taking everything you already know and re-mapping it onto a new machine that does the same job with a different cockpit. Starting a new SRE job is the same thing.

At the end of last year, our uptime was pretty shaky. Many of these outages were caused by a single bug, deep in SQLite. It took months of intense forensics to track it down.

Pod distribution plays a much bigger role in reliability than you might think. By adding a few lines to your manifest, you can ensure your deployments are zone-redundant and evenly scalable. The feature is called topology spread constraints, and in this blog, we'll explain how it works in full detail.

Rune is a fast, GPU-accelerated, full-featured IDE and terminal multiplexer, suitable both for automatic and manual programming.

Linux sandboxing that doesn't get in your way

High-performance code-intelligence engine for AI agents and IDE, supports 257 languages, multi repositories, based on graph, with access via CLI, MCP Server, and API. AI coding agents teammate - expose only needed information, cutting token usage up to 50x. 100% local.

Atlas is source control for coding agents. Every agent run produces checkpoints: commits are linked back to the session that made it alongside the prompts, tool calls, and reasoning. You see which agent did exactly what and why.

Run Claude Code, Codex, Atlas's own agent, or anything from the ACP registry side by side against the same codebase, with shared memory so switching agents mid-task doesn't mean starting over.

Declare an agentic task with workspaces and gateway specifications. AX sandboxes it, wires up its workspace, fences its network, and helps running it at scale.

AX is a high-throughput, declarative orchestrator to run billions of autonomous agent workloads in a cluster. It runs on top of Agent Substrate for sandboxed execution and is built to run billions of tasks per cluster. If you have used Kubernetes, ax will feel similar.

Distributed, offline-first bug tracker embedded in git

When the scripts that generate the data for letsencrypt.org/stats broke yet again, we decided to retire it rather than repair it. Let’s Encrypt issues six to ten million certificates each day, producing a large volume of logs that keeps growing. It became increasingly time-consuming and difficult to answer questions about our own issuance like “how many certificates use the ‘shortlived’ profile.” Using raw logs, this requires finding, parsing and extracting relevant portions of loglines. Querying the database behind our issuance API is not a practical option, as it’s built for transactions rather than analysis. We’d also used a log search SaaS product, but our bills were growing much faster than we’d like, and while it was fine for searching, it wasn’t able to do the analytic workloads we needed. We knew we could dream bigger and better.

walgit with a GitHub Enterprise Server facade — git over smart HTTP from an object-store bucket

Whiteboard is an open-source desktop app where humans and agents can architect software together in a common workspace.

A harness wraps a model. A rig wraps your harnesses. Define your agent team in YAML, boot it with one command. Claude Code and Codex in the same rig, managed as one system.

OpenRig turns AI coding agents from a pile of terminal sessions into a persistent, organized team. Talk to a lead agent about the outcome you want; it can coordinate specialists across teams and bring you results and decisions that need your attention. Start with a repository and one useful change, then keep the team's work and context at the same addresses.

As a part of the journey to transition Netflix's compute infrastructure to be more Kubernetes-native, we have leaned into incorporating components from the Kubernetes ecosystem into our container platform Titus. One example of this is our use of Kueue, a cloud-native job queueing system for batch workloads, which has largely replaced the custom queuing and scheduling logic in our homegrown managed batch solution Compute Managed Batch (CMB).

Server-side apply matters because Kubernetes objects are shared state: it moves field ownership into the API server, so apply-style tools can surface conflicts instead of hiding them as silent overwrites.

I maintain an app that builds on top of kubernetes port forwarding, so i track KEP-4006 because the streaming protocol underneath keeps changing. I wrote about this back in april 2024 around Kubernetes 1.30, and six releases later it's changed enough to be worth another look.

You've got ten Claude Code agents running. Each is working on a different issue. Agent 3 just created a PR that conflicts with Agent 7's changes. Agent 5 has been stuck on "Analyzing codebase…" for 45 minutes. Welcome to multi-agent chaos. Turns out, scaling AI coding agents from 1 to 10+ is a completely different problem from running a single agent in a terminal.