# Shopify Engineering - Shopify Engineering Stories and resources from the teams who build and scale Shopify, the leading cloud-based, multi-channel commerce platform. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Migrating Shop app from React Native to native DevFeed: [Migrating Shop app from React Native to native]() Original publisher: [Read original article]() Author: Max Da Silva Published: 2026-09-10T12:27:06Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [migration]() Tags: [agents](), [ai](), [android](), [compose](), [ios](), [kotlin](), [migration](), [react-native](), [swift]() ### AI overview Shopify describes migrating the Shop app from React Native to native Swift and Kotlin development. Coding agents supported a proof of concept and the subsequent rebuild while the team aimed to preserve feature behavior and analytics events. ### Source excerpt We migrated the Shop app from React Native to Swift and Kotlin. Assisted by AI, the team was able to go from a proof of concept to a fully rebuilt native app published in the app stores in just 12 weeks. ## Native is now the future of mobile at Shopify DevFeed: [Native is now the future of mobile at Shopify]() Original publisher: [Read original article]() Author: Mustafa Ali Published: 2026-09-10T12:25:28Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [React Native](), [AI-assisted coding](), [Large Language Model](), [iOS]() Tags: [agents](), [android](), [apps](), [coding](), [kotlin](), [llms](), [mobile](), [models](), [react-native](), [shopify](), [swift]() ### AI overview Shopify is moving from React Native back to native Swift and Kotlin development after finding that coding agents and LLMs reduce the cost of implementing mobile features separately on each platform. ### Source excerpt Coding agents changed what it costs to build mobile apps twice. Here's why Shopify is moving from React Native back to Swift and Kotlin. ## How River takes security work from a fix to merge DevFeed: [How River takes security work from a fix to merge]() Original publisher: [Read original article]() Author: Erin Son Published: 2026-09-02T16:30:45Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Vulnerabilities](), [AI Bots]() Tags: [agentic](), [ai-agent](), [autonomous](), [ci](), [security](), [shopify](), [slack](), [vulnerability](), [workflow]() ### AI overview Shopify describes River, an AI agent that manages vulnerability remediation from patch review through merge and verification against the repository head and dependency graph. The article reports a 70% reduction in open dependency issues during the workflow's first 11 days. ### Source excerpt River, Shopify's AI agent in Slack, enables autonomous vulnerability remediation, not just detection. ## Gisting: Compressing LLM Agent context to ↑ throughput and ↓ cost DevFeed: [Gisting: Compressing LLM Agent context to ↑ throughput and ↓ cost]() Original publisher: [Read original article]() Author: Cody Mazza-Anthony Published: 2026-08-19T14:32:58Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Compression](), [Large Language Model](), [Embeddings](), [Latency](), [Inference](), [Inference Performance](), [Low-Latency Inference](), [Post-training optimization](), [GraphQL](), [model-deployment]() Tags: [compression](), [embedding](), [inference](), [latency](), [llm](), [model](), [tokens]() ### AI overview Gisting compresses an LLM agent's system prompt into learned gist tokens, preserving prediction quality while reducing inference latency, increasing throughput, and lowering GPU requirements. ### Source excerpt Gisting compresses context into a set of learned tokens, preserving its quality while making the model faster and cheaper. ## How we raised mobile end-to-end test stability to 98% DevFeed: [How we raised mobile end-to-end test stability to 98%]() Original publisher: [Read original article]() Author: Michael Garfinkle Published: 2026-08-12T20:02:35Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [ci](), [React Native](), [App](), [Pull Request](), [User Experience]() Tags: [app](), [ci](), [computer-vision](), [mobile](), [pull-request](), [react-native](), [testing](), [user-experience]() ### AI overview Shopify rebuilt its mobile end-to-end testing framework with a strict builder-style API and computer vision. The approach raised test stability from 50% to 98% and restored reliable blocking checks for pull requests. ### Source excerpt We rebuilt our mobile end-to-end testing framework with a strict API and computer vision, raising test stability drastically. ## Sidekick's continual learning loop DevFeed: [Sidekick's continual learning loop]() Original publisher: [Read original article]() Author: Andrew McNamara Published: 2026-08-05T14:52:54Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Artificial Intelligence](), [Shopify](), [GraphQL](), [Ground truth / benchmark quality](), [Latency](), [Code]() Tags: [agent](), [ai](), [frontier-model](), [latency](), [model](), [production](), [retrieval](), [routing](), [safety]() ### AI overview Shopify describes a continual learning loop for its GraphQL agent that turns production failures, user corrections, and sampled traffic into ground truth and model-weight improvements. The approach is reported to exceed frontier-model quality while reducing latency and serving costs by 96%. ### Source excerpt How we compress production failures into model weights every day, beat frontier-model quality, and cut serving costs 96%. ## Building an agentic harness that outlasts the model DevFeed: [Building an agentic harness that outlasts the model]() Original publisher: [Read original article]() Author: Zack Deveau Published: 2026-07-29T15:30:00Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Application Security](), [Shopify](), [Vulnerabilities](), [Code review](), [Orchestration](), [pull-requests](), [Rails](), [Ruby](), [web applications](), [Back end]() Tags: [agentic](), [application-security](), [backend](), [building](), [code](), [orchestration](), [pull-requests](), [review](), [ruby](), [shopify](), [vulnerabilities](), [web-applications](), [workflow]() ### AI overview Shopify describes an agentic code review and test-oracle harness for application security. The harness scans software for vulnerabilities, validates findings with real tests, generates Shopify-specific fixes, and opens relevant pull requests. The article also explains its Dispatch orchestrator, parallel scanning workflow, reusable application context, and diff-based follow-up scans. ### Source excerpt We built an agentic code review and test oracle harness that discovers vulnerabilities, proves them with real tests, and provides Shopify-tuned fixes. ## Upgrading Checkout Blocks app to Polaris web components DevFeed: [Upgrading Checkout Blocks app to Polaris web components]() Original publisher: [Read original article]() Author: Justin Henricks Published: 2026-07-16T17:19:40Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Latency](), [React](), [GraphQL](), [telemetry](), [API]() Tags: [api](), [apis](), [architecture](), [checkout](), [latency](), [performance](), [react](), [telemetry](), [typescript](), [web]() ### AI overview Checkout Blocks migrated five high-traffic checkout extensions from Shopify's legacy Remote UI and React-based path to remote-dom, Polaris web components, Preact, and TypeScript. The upgrade reduced transferred bundle sizes by 40% to 85% and improved extension loading performance. ### Source excerpt We moved five high-traffic checkout extensions to remote-dom and Polaris web components, cutting bundle sizes drastically and making checkout faster. ## Inside Shopify Hack Days: Building a prototype for music-playing pages DevFeed: [Inside Shopify Hack Days: Building a prototype for music-playing pages]() Original publisher: [Read original article]() Author: Justin Henricks Published: 2026-07-14T13:55:57Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [glsl](), [real-time](), [App](), [Web](), [API](), [Template]() Tags: [3d](), [api](), [app](), [audio](), [generation](), [glsl](), [hackathon](), [interactive-3d](), [real-time](), [shopify](), [web]() ### AI overview Shopify's Hack Days team built a prototype for music-playing product pages in three days. The project combined a Shopify app, audio playback, per-track purchasing, GLSL visualizers driven by real-time audio, synced lyrics, and tour-date information. ### Source excerpt How a Hack Days team built a music player, custom GLSL visualizers, and an artist toolkit for storefronts, all in three days. ## Clustering billions of products for agentic commerce with Catalog API DevFeed: [Clustering billions of products for agentic commerce with Catalog API]() Original publisher: [Read original article]() Author: Mariya Mansurova Published: 2026-06-17T16:54:51Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [AI search](), [AI Bots](), [SDKs]() Tags: [agentic](), [agents](), [ai](), [api](), [data](), [developers](), [llm](), [products](), [search](), [shopify]() ### AI overview Shopify describes an LLM-powered product-clustering pipeline that reconciles differently structured merchant listings into universal product groups for AI-agent search through the Catalog API. ### Source excerpt Building an LLM-powered pipeline to match product listings across millions of merchants into a unified catalog that AI agents can search. ## Teaching Sidekick to say no: automated data curation with LLM judge consensus DevFeed: [Teaching Sidekick to say no: automated data curation with LLM judge consensus]() Original publisher: [Read original article]() Author: Shuang Xie Published: 2026-06-15T19:30:00Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [data](), [Large Language Model](), [Fine-tuning](), [AI Development](), [foundation-models](), [Training AI Models](), [Artificial Intelligence](), [Open Source Models & Datasets](), [Open Source]() Tags: [agentic](), [ai](), [analysis](), [customer](), [data](), [feature-engineering](), [fine-tuning](), [llm](), [open-source](), [shopify](), [training]() ### AI overview Shopify Engineering describes how Sidekick's production training data failed to teach refusal behavior because it contained only successful merchant queries. The article presents automated data curation using consensus among LLM judges to identify blind spots and improve an AI assistant built from an outer planner and specialized skill models. ### Source excerpt Production training data only captures successful queries; it can't teach a model when to say no. We built an automated curation pipeline using LLM judge consensus to close that gap. ## Quick: An internal hosting platform for the AI era DevFeed: [Quick: An internal hosting platform for the AI era]() Original publisher: [Read original article]() Author: Daniel Beauchamp Published: 2026-06-10T12:49:23Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [hosting](), [Shopify](), [Artificial Intelligence](), [API](), [Cloud](), [nginx](), [HTML](), [Filesystems](), [Web Development]() Tags: [ai](), [api](), [cloud](), [hosting](), [html](), [server](), [shopify](), [storage]() ### AI overview Shopify's Quick is an internal hosting platform that lets employees publish folders of HTML and assets to secure, employee-only URLs without frameworks, deployment pipelines, or configuration files. The article describes its Google Cloud Storage and NGINX architecture, Identity-Aware Proxy authentication, local-folder upload workflow, and planned API capabilities for data, file storage, and AI-related functionality. ### Source excerpt Quick lets anyone at Shopify ship a site in seconds. It has changed the culture of how we build and share. ## How Shopify Built the Infrastructure Beneath Its Slack-Native River AI Agent DevFeed: [How Shopify Built the Infrastructure Beneath Its Slack-Native River AI Agent]() Original publisher: [Read original article]() Author: Burke Libbey Published: 2026-05-28T12:30:02Z Content type: opinion Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [AI Agent](), [Slack](), [monorepo](), [Nix](), [Development](), [pull-requests]() Tags: [agent](), [ai-agent](), [ai-coding](), [cache](), [monorepo](), [slack](), [test]() ### AI overview Shopify describes how it built River, an AI agent that operates in company Slack, and the infrastructure choices that supported it. The article covers Shopify's move to a monorepo, adoption of Nix for reproducible environments, and the resulting changes to CI, merge queues, build caching, testing, and repository navigation. ### Source excerpt What it took to ship our Slack-native agent River, lessons learned, and the substrate that runs beneath it. Co-authored by River. ## We replaced Redis with MySQL for inventory reservations--and it scaled DevFeed: [We replaced Redis with MySQL for inventory reservations--and it scaled]() Original publisher: [Read original article]() Author: Emilie Noel Published: 2026-05-12T16:51:53Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [MySQL](), [Redis](), [Database](), [Shopify](), [Transactions]() Tags: [database](), [ecommerce](), [mysql](), [payment-processing](), [performance](), [redis](), [scale](), [shopify](), [transactions]() ### AI overview Shopify explains how it replaced Redis with MySQL for inventory reservations during checkout. Using one row per inventory unit, SKIP LOCKED, composite primary keys, and connection visibility, the rebuilt system met high-throughput and consistency requirements during peak Black Friday 2025 traffic. ### Source excerpt How we used SKIP LOCKED, composite primary keys, and connection visibility to hit our scale targets. ## Flow generation through natural language: An agentic modeling approach DevFeed: [Flow generation through natural language: An agentic modeling approach]() Original publisher: [Read original article]() Author: Ted Chaiwachirasak Published: 2026-04-22T12:32:02Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [AI Bots](), [AI Chat]() Tags: [agent](), [agentic](), [ai](), [automation](), [benchmark](), [fine-tuning](), [model](), [synthetic](), [training](), [workflow]() ### AI overview Shopify describes fine-tuning Qwen3-32B into a tool-calling agent that generates Flow automations from natural-language requests. The article focuses on constructing synthetic training data from validated workflows and evaluating the resulting model. ### Source excerpt We fine-tuned Qwen3-32B into a tool-calling agent that generates Flow automations from natural language--faster, cheaper, and more accurate than the frontier model it replaced, with a weekly retraining flywheel built on real merchant data. ## Autoresearch isn't just for training models DevFeed: [Autoresearch isn't just for training models]() Original publisher: [Read original article]() Author: David Cortés Published: 2026-04-15T16:19:20Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [AI Bots](), [CI/CD]() Tags: [agent](), [ai](), [ci](), [metrics](), [models](), [performance](), [shopify](), [training]() ### AI overview The article describes applying an AI-driven Autoresearch loop to improve engineering metrics, beginning with efforts to reduce CI and test execution time. ### Source excerpt Tobi and I generalized Karpathy's Autoresearch to improve 40+ metrics across Shopify, then we open-sourced our project. ## Building a Magic Mirror: AI retail experiences with Remix DevFeed: [Building a Magic Mirror: AI retail experiences with Remix]() Original publisher: [Read original article]() Author: Nikola Draca Published: 2026-03-19T12:33:46Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Artificial Intelligence](), [Remix](), [webcam](), [Shopify](), [browser](), [API]() Tags: [ai](), [api](), [browser](), [building](), [code](), [customer](), [generate](), [how-to](), [mirror](), [product](), [retail](), [server](), [shopify](), [webcam]() ### AI overview This article presents Shopify's AI-powered magic mirror: a customizable retail installation that uses a display, hidden webcam, and Remix server to recognize visual signals and deliver personalized messages, animations, product recommendations, challenges, and discount codes. It also outlines a makeup shade-matching use case and the hardware needed to build the experience. ### Source excerpt The technical blueprint for an AI-powered mirror that sees customers, analyzes their appearance, and delivers personalized product recommendations in real-time. ## Shopify's journey to faster breadth-first GraphQL execution DevFeed: [Shopify's journey to faster breadth-first GraphQL execution]() Original publisher: [Read original article]() Author: Greg MacWilliam Published: 2026-03-12T12:19:08Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [GraphQL](), [Shopify](), [data](), [Traces]() Tags: [apis](), [data](), [shopify](), [technology](), [traces]() ### AI overview Shopify describes how its GraphQL Cardinal engine replaces conventional depth-first execution with breadth-first field resolution. The approach addresses resolver overhead in deeply nested, high-cardinality queries and reportedly delivers up to 15x faster execution with 90% less memory use for large list queries. ### Source excerpt Conventional GraphQL execution uses depth-first traversal that incurs many hidden costs. We questioned why, rewrote it in a faster breadth-first manner, and saw dramatic results. ## 2,000 robots walk into a shop... DevFeed: [2,000 robots walk into a shop...]() Original publisher: [Read original article]() Author: Javier Moreno Published: 2026-02-27T18:11:45Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [AI Bots](), [AI Chat]() Tags: [browser](), [chromium](), [cloud](), [collaboration](), [cost](), [gpt](), [inference](), [infrastructure](), [json](), [latency](), [llm](), [nvidia](), [robots](), [shopify](), [testing]() ### AI overview Shopify describes SimGym, a system that uses LLM-guided browser robots to simulate shopper behavior and compare storefront themes. The article focuses on the infrastructure and latency trade-offs of running many cloud-browser sessions alongside model inference. ### Source excerpt SimGym infrastructure for synthetic customers at scale. A Shopify + NVIDIA collaboration. ## The generative recommender behind Shopify's commerce engine DevFeed: [The generative recommender behind Shopify's commerce engine]() Original publisher: [Read original article]() Author: Yang Liu Published: 2026-02-25T16:04:54Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Shopify](), [Temporal data](), [model-deployment](), [real-time](), [Feature Engineering]() Tags: [architecture](), [engineering](), [feature-engineering](), [generative](), [latency](), [model](), [production](), [real-time](), [scale](), [shopify](), [train](), [training]() ### AI overview Shopify describes a generative recommender that treats buyer journeys as raw event sequences. An autoregressive, causally masked model predicts next products or ads while meeting real-time production constraints at Shopify's scale. ### Source excerpt Treating buyer journeys as sequences instead of simplified signals, and building a model fast enough to serve at scale. ## How Shopify uses SkyPilot to route machine-learning workloads across multi-cloud GPU clusters DevFeed: [How Shopify uses SkyPilot to route machine-learning workloads across multi-cloud GPU clusters]() Original publisher: [Read original article]() Author: Javier Moreno Published: 2026-01-26T14:49:55Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [skypilot](), [Cloud](), [Machine learning](), [InfiniBand]() Tags: [cloud](), [development](), [distributed-training](), [infiniband](), [kubernetes](), [kubernetes-clusters](), [multi-cloud](), [open-source](), [shopify](), [skypilot](), [yaml]() ### AI overview Shopify describes using SkyPilot to run machine-learning workloads across existing Kubernetes clusters on multiple clouds. A custom plugin routes jobs based on requested hardware and workload needs, while supporting multi-team management, cost tracking, fair scheduling, and policy enforcement. ### Source excerpt GPUs are annoying. Shopify uses SkyPilot to make them less so: one YAML file, multiple clouds, clean development ergonomics. ## Building the Universal Commerce Protocol DevFeed: [Building the Universal Commerce Protocol]() Original publisher: [Read original article]() Author: Ilya Grigorik Published: 2026-01-11T14:17:01Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [AI Bots]() Tags: [agentic](), [ai-agents](), [apis](), [architecture](), [extensions]() ### AI overview Shopify describes UCP, an open protocol for AI agents to discover merchant capabilities, negotiate supported actions, and complete commerce transactions. ### Source excerpt We built UCP to power agentic commerce. Here's the architecture behind it. ## Tangle: An open-source ML experimentation platform built at Shopify scale DevFeed: [Tangle: An open-source ML experimentation platform built at Shopify scale]() Original publisher: [Read original article]() Author: Shopify Engineering Published: 2025-12-05T15:47:00Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Machine Learning & Artificial Intelligence](), [Open Source](), [Shopify](), [data-engineering](), [experiments]() Tags: [caching](), [cloud](), [collaboration](), [data-engineering](), [experimentation](), [ml](), [open-source](), [pipelines](), [shopify]() ### AI overview Shopify open-sourced Tangle, an ML experimentation platform designed to make experiments reproducible, accelerate iteration, and enable teams to share pipelines and computation. It supports visual ML and data pipelines, cloud execution, and global caching, reducing repeated data preparation, custom pipeline maintenance, and infrastructure costs. ### Source excerpt Tangle saves months of compute time, makes every experiment automatically reproducible, and allows teammates to share computation without coordination. ## Behind the build: 2025's BFCM globe in a pinball machine 🌎🕹 DevFeed: [Behind the build: 2025's BFCM globe in a pinball machine 🌎🕹]() Original publisher: [Read original article]() Author: Shopify Engineering Published: 2025-12-03T14:28:24Z Content type: article Language: en Sources: [Shopify Engineering](), [Shopify Engineering - Shopify Engineering]() Topics: [Shopify](), [Three.js](), [real-time](), [React](), [browser](), [Three.js shaders]() Tags: [2025](), [3d](), [browser](), [js](), [performance](), [physics](), [pixel](), [react](), [react-three-fiber](), [real-time](), [server-sent-events](), [shopify](), [stream](), [three-js]() ### AI overview Shopify describes its 2025 BFCM live globe, implemented as a browser-based pinball machine that visualizes merchant sales in real time. The three.js app uses react-three-fiber, server-sent events, a physics engine, VR support, and shaders for its display animations. ### Source excerpt Real-time merchant sales rendered on our globe, Easter eggs galore, and a pinball machine because... why not? [Next page]()