Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
Animations for System Builders: Metastability and Lost Writes
Marc Brooker shows how LLM-built animations backed by real test traces help build intuition for distributed-systems failures, including how a Raft cluster can lose writes when one node's storage lies, and metastable failure.
2 min · 509 words
SBD: We Have a Filesystem at Home
Fly.io's Daniel Botha introduces the Sprites Block Device (SBD), a simpler object-storage-backed block device that replaces the clever custom filesystem Sprites launched with, explaining why dumber checkpoints and 'just a disk' made Sprites more reliable.
6 min · 1,282 words
Evolving the cloud-native foundation for the next generation of AI
Microsoft Azure describes how it is advancing cloud-native infrastructure so organizations can build, run, and optimize applications and agents at scale for the next generation of AI workloads.
7 min · 1,552 words
Postgres DELETE Can Actually Scale
DBOS explains how Postgres actually deletes rows and how it made workflow retention deletes scale: batched deletes of the oldest workflows, dropping foreign keys so tables are cleaned in phases, debugging a 20-minute index lookup, and why it chose DELETE over dropping partitions.
7 min · 1,532 words
What changes when your search index lives on S3?
Infino explains how putting a full-text search index on object storage changes the cost and latency model: every random read is a billed GET taking tens of milliseconds, and how superfiles, manifests and careful layout keep agent queries fast without packing the working set on local disks.
9 min · 2,019 words
How the Claude Code team does code review
With Claude writing about 90 percent of Claude Code's own code, the team describes a layered review flow: verification before review, AI review on every pull request, blocking security and automated checks, risk-scaled human approval, merge queues, flags, gradual release and fast rollback. Merged PRs more than doubled from June to September while the revert share fell over 40 percent.
15 min · 3,429 words
How we built AI agent behavior drift analysis with Cloudflare Clef
David Mytton explains how Arcjet monitors AI coding agent sessions in real time, detecting risky actions, session tainting and behavior drift from indirect prompt injection using Cloudflare's Clef decision model.
9 min · 1,997 words
E2exe: How we solved Windows desktop automation for enterprise scale
QA Wolf explains how it parallelizes Windows desktop app tests at scale: single-use VMs from warm pools on Google Compute Engine, driven from Linux over CDP and Appium, with one API across web, mobile and desktop.
10 min · 2,225 words
Inside Slipstream’s Recall Pattern
Glide’s engineering team explains Slipstream’s recall pattern for handing six types of rote engineering maintenance work to AI agents: how issues are detected, recalled into agent workflows, verified, and measured, with the numbers they are seeing in production.
9 min · 2,016 words
Building a Linearizable Log on READ COMMITTED Isolation
Materialize describes how a metastable incident under PostgreSQL SERIALIZABLE led them to build a linearizable metadata log on READ COMMITTED isolation, achieving 3x write throughput, backed by a Lean 4 proof.
20 min · 4,657 words
Rewinding the exponential: How we keep hardware usage down
Cloudflare’s Data team explains how it reined in an exponentially growing analytics pipeline—rewinding retention, compression, and query patterns—so hardware and cost stay sustainable while more than 20% of the web sits behind its network.
10 min · 2,215 words
Redesigning source control at Anthropic
Anthropic explains how merges to its monorepo grew from about 2,500 to over 30,000 per day in 2026, why git hosting became the bottleneck, and how it built and migrated to Scion, an in-house version control system, in five months, plus lessons and ideas for the wider git ecosystem.
16 min · 3,745 words
How we built demo.usenotra.com
Notra founding engineer Jan Burzinski explains how every visitor to demo.usenotra.com gets an isolated workspace in about 0.4 seconds without signing up—using signed cookies, a Postgres sandbox pool, and a fake model—so people can try the product immediately.
8 min · 1,924 words
Tetris in d dimensions: how we rightsize Kubernetes nodes
Emmanuel Bakare explains how CostGraph rightsizes Kubernetes nodes as a vector bin packing problem: sizing pods from real usage percentiles, checking placements against scheduler hard constraints, and using Dot Product and L2 heuristics from Panigrahy et al. instead of first fit decreasing.
12 min · 2,675 words
Speeding up the Unreal editor by… not using unnecessarily secure hash methods
A profiling write-up finding that Unreal Engine's editor launch hashes full shader source files with cryptographically secure SHA-1, and how switching to the engine's existing xxHash implementation sped this up, plus notes on preparing the change as a pull request.
4 min · 986 words
Migrating a banking app from AngularJS to Web Components
Marius Vatasoiu recounts the six-plus-year migration of ING Bank Romania's feature-rich banking app from AngularJS to Web Components with Lit, done incrementally without impacting clients, and the lessons learned about architecture, routing and working through framework abstractions.
12 min · 2,858 words
When ldaxr Doesn't Work: Exclusive Accesses and Cacheability on AArch64
Joel Siks debugs why spinlocks built on ldaxr/stxr exclusive accesses hung on a real Raspberry Pi 5 while working in QEMU in his hobby AArch64 kernel floss, tracing it to memory cacheability attributes and the differences between QEMU, the Arm architecture manual and the Cortex-A76.
12 min · 2,658 words
We gave init() developers an ESP32-S3 computer as a conference badge. They put DOOM on it.
WorkOS shipped ESP32-S3 computers as init() 2026 conference badges; attendees put DOOM, Tamagotchi, voice assistants, and flight trackers on them. A recap of the badge hardware and what ~500 people built.
5 min · 1,195 words
Daniel Mangum digs into the Nordic nRF54LM20 access port and what you can learn when a peripheral writes memory while the CPU is halted.
9 min · 1,960 words
Scaling and benchmarking a critical message bus using a new indexing strategy
Jane Street describes a summer intern project on Aria, its internal messaging framework: indexing the in-memory stream tip and adaptively splitting the topic tree so clients can cheaply recover just the messages they need, cutting CPU usage by about 30% on production-like workloads.
9 min · 2,000 words