Stripe
Modeling Money Movement: Ledgers, State Machines, and Exactly-Once Effects
How Stripe models money movement with immutable ledger entries and explicit state machines to guarantee correctness even when networks and services fail.
3 min read
Tag
Engineering at Scale
96 articles
Spotify
From Peer-to-Peer to CDNs: How Spotify Actually Streams Audio
Why Spotify's early client-side streaming relied on peer-to-peer delivery, and why it later moved entirely to server and CDN-based delivery instead.
3 min read
Amazon
Why Amazon Bet Early on Queues: The Philosophy Behind SQS
How Amazon's early commitment to asynchronous, queue-based decoupling shaped SQS and became a default pattern across its service architecture.
3 min read
Airbnb
From Heuristics to Learned Ranking: Airbnb Search's Evolution
How Airbnb's search ranking moved from hand-tuned heuristics to gradient-boosted trees and eventually neural networks, and what each transition actually bought.
3 min read
Amazon
Two Pizzas, One Team: Amazon's Operational Culture Beyond the Slogan
How Amazon's two-pizza teams, operational readiness reviews, and correction-of-error documents work together as an operating system, not just slogans.
3 min read
Meta
MyRocks: How Facebook Cut MySQL Storage Nearly in Half
Why Facebook replaced InnoDB with MyRocks, an LSM-tree storage engine built on RocksDB, to shrink storage footprint across its huge MySQL fleet.
3 min read
Google
Dremel: How Google Made Trillion-Row Queries Feel Interactive
The columnar storage and execution-tree design behind Google's Dremel engine, and how it became the foundation for BigQuery.
3 min read
Netflix
Chaos Engineering at Netflix: Breaking Things on Purpose
How a painful database outage pushed Netflix to build Chaos Monkey, the Simian Army, and a culture that tests failure before failure finds you.
3 min read
Stripe
Sorbet: Bringing Static Types to Millions of Lines of Stripe's Ruby
Why Stripe built its own type checker, Sorbet, instead of rewriting its large Ruby codebase, and how gradual typing let adoption happen incrementally.
3 min read
Netflix
Netflix's Homepage Isn't One Recommender — It's Dozens Working Together
Why Netflix's personalization system is architected as many specialized algorithms for ranking, rows, and artwork rather than a single model.
3 min read
Pinterest
Why Pinterest Bet on Sharded MySQL Over NoSQL
Pinterest hit early scaling walls, tried NoSQL alternatives, then built ID-encoded sharded MySQL instead — and it outlasted the exotic options.
3 min read
Spotify
From Log Files to Cloud Pub/Sub: Spotify's Event Delivery Evolution
How Spotify's event pipeline evolved from batch log collection off servers to a real-time, cloud-native Pub/Sub system serving hundreds of consuming teams.
2 min read
Amazon
Aurora: What Happens When You Rebuild the Database Around the Log
How Amazon Aurora separates database compute from a purpose-built, log-structured distributed storage layer to cut replication network traffic.
3 min read
Airbnb
Airbnb's Design System: Component-Driven Frontend at Scale
How Airbnb replaced inconsistent, hand-rolled UI across its apps with a shared design language and component library engineers and designers could both trust.
3 min read
Uber
Cadence: Writing Workflow Code as If Failures Never Happen
How Uber's Cadence engine lets developers write long-running workflow logic as plain code while the platform handles retries and state durably.
3 min read
DoorDash
DoorDash's Break From the Monolith: A Platform-First Migration
How DoorDash extracted its logistics and dispatch systems from a Django monolith first, and built a service platform to avoid microservices sprawl.
3 min read
Meta
The Origin of React: UI as a Function of State
How Facebook's Ads team, wrestling with cascading manual DOM updates, built React around declarative rendering and a virtual DOM diff.
3 min read
WhatsApp
How WhatsApp Served Millions of Connections per Server With a Tiny Team
How Erlang's lightweight process model and deep FreeBSD tuning let WhatsApp run hundreds of millions of users' traffic with a remarkably small team.
3 min read
Google
Error Budgets and Toil: How Google Turned Operations Into an Engineering Discipline
How Google's Site Reliability Engineering practice used error budgets and a hard cap on toil to align reliability work with product velocity.
3 min read
Stripe
Stripe's Four-Phase Pattern for Migrating Databases Without Downtime
How Stripe moves data between database schemas and systems while staying live, using a four-phase dual-write and backfill pattern instead of a maintenance window.
3 min read
Pinterest
The Feature Infrastructure Behind Pinterest's Home Feed Ranking
How Pinterest built shared machine learning feature infrastructure to rank home feed content consistently across training and real-time serving.
3 min read
Uber
Michelangelo: How Uber Turned Machine Learning Into a Self-Service Platform
How Uber built Michelangelo to give data scientists a shared path from feature data to a deployed, monitored production ML model.
3 min read
Spotify
Backstage: How Spotify's Internal Sprawl Led to a Developer Portal
How thousands of undiscoverable microservices inside Spotify led to Backstage, the developer portal it later open sourced through the CNCF.
2 min read
DoorDash
Reliability Engineering for a Three-Sided Marketplace
How DoorDash designs for graceful degradation and load shedding so a spike or partial outage on one side of its marketplace doesn't take down the rest.
3 min read