[1/2] How AI is changing software engineering at Shopify (AI column)

I saw something interesting. interview Gergely Orosz and Farhan Thawar, who directs the engineering department of Shopify, overseeing more 3 000 engineers. Prior to Shopify, he worked at Pivotal Labs, Xtreme Labs and co-founded Helpful.com, which Shopify acquired. Known for his hands-on approach to leadership - for example, in an interview, he recounted how personally fixing Wi-Fi at a corporate event:). Gergely Orosz is the author of The Pragmatic Engineer, author and conference speaker (recently told “Software Engineering with LLMs in” 2025: reality check").

The interview was interesting and his key ideas are presented below.

1. Shopify philosophy: pair programming problems Shopify takes a “hire smart people and work with them on problems” approach rather than the traditional “hire and leave them alone.” Tawar emphasizes that the company does not divide employees by roles - everyone expects to solve problems regardless of position. (You can't just delegate it down.). 2. Early implementation of AI tools Shopify was one of the first companies to use GitHub Copilot. In 2021 Prior to the launch of ChatGPT, Tawar personally approached the new CEO of GitHub to request access to Copilot for all Shopify engineers. The company received free access for two years in exchange for feedback. 3. Diversification of AI Tools Although Shopify traditionally prefers one tool for each task, with the advent of AI, the company changed its approach and began to use several AI tools to avoid putting all its eggs in one basket. GitHub Copilot and Cursor for development Claude Code for Agent Workflows

• Devin (tested) 4. AI beyond development Cursor is actively used not only by engineers, but also by finance, sales and support teams. Non-technical employees create MCP servers to access Salesforce, Google Calendar, Gmail, and Slack, often without the help of engineers. 5. "Code Red" - the fight against technical debt Shopify has held 7- Monthly Operation Code Red. (November 2024 - June 2025)directed 30-50Percentage of engineers to eliminate technical debt 6. Memorandum by CEO Shopify on AI Shopify CEO Toby Lutke publishes internal memorandumThe use of AI is mandatory for all employees. Key provisions: Reflex use of AI is a basic expectation Before requesting additional resources, it is necessary to prove that the AI cannot perform the task. Competence in AI will be part of performance assessment 7. AI infrastructure in Shopify
• LLM proxy: Internal proxy server for secure use of language models without leaking corporate data. Includes token tracking and a leaderboard of users with the highest consumption.
• MCP servers: More than two dozen Model Context Protocol servers provide access to internal data, including the corporate wiki "The Vault" with project information and transcripts of presentations. 8. Philosophy of spending on AI Shopify does not limit spending on AI tokens. Tawar claims that even $1000 a month for an engineer - "too cheap" for tools that increase productivity 10%. The company encourages the use of advanced models instead of basic ones. 9. Internship programme Shopify plans to hire 1000 trainee 2025 They call them "AI centaurs" -- people who work effectively with language models. Internships last 4 months and require an office presence (Toronto, Ottawa, Montreal) To create a culture of collective learning.

Conclusions from the report in continuation.

#AI #ML #PlatformEngineering #Software #Architecture #Processes #DevEx #Devops