What candidates say the round looked like, summarised in our words, each linked to where it came from. Shopify hasn’t confirmed these details unless the source says “Official page”.
Had this round? Add your report →
This intern candidate pasted the problem statement into Claude, asked it to plan the implementation, approved the plan, and had it implement and test. Their review found nothing to change, and the one follow-up feature was also trivial with Claude. Their bigger advice was about the human side: be an active listener, include the interviewer in the review, and build a genuine connection.
Intern pair programming round, 45 minutes, with you as the driver and AI tools explicitly encouraged. A previous candidate said to use AI for everything, from ideation and planning to implementation and tests, while showing you don’t blindly accept it; in their experience using it was close to mandatory to pass. The problem was easy: design a small class with 2 or 3 simple methods plus tests, guarding against impossible operations, with follow-ups in the same vein. The GitHub repo was empty and only used to push your code, so check your git credentials beforehand.
Staff-level first technical round under Shopify’s bring your own everything model: your own IDE with your AI assistants set up before the call. Over 70 minutes the candidate finished both iterations of a known Shopify-tagged problem, prompting the AI to build on a boilerplate HTTP server, then reviewing, adjusting and explaining the code. Asked what else they’d add, they covered error handling, proper 400/500 responses, telemetry, metrics, API versioning and an external cache. They were rejected the next day, and said AI rounds make it hard to tell what was expected.
About 20 minutes went into designing the solution, which the candidate felt went well, and then handing the design to AI to write the code and refining it. They had no idea afterwards what had been expected. Another candidate in the SQL + AI version failed and concluded Shopify still wants to see that you can implement things yourself as well as guide AI.