Shopify: How AI is changing software development | Featuring Farhan Thawar, VP Engineering

If you’ve ever wondered how Shopify moves so fast, much of that DNA traces back to Farhan Thawar, Shopify’s Head of Engineering.

Farhan has spent his career building high-velocity engineering teams, from Pivotal Labs to Xtreme Labs to Shopify. His worldview is simple but profound: most leaders are drowning in unnecessary complexity, and most teams are slowed down by invisible friction. He believes that if you want to do truly hard, high-leverage work, the path to get there is to make your jobs as easy as possible, freeing up time for the juicy (hard!) stuff.

At Shopify, he’s known for championing a culture of speed, craftsmanship, and constant simplification, and for some bold organizational experiments with his team.

If you’d been there, here’s what you’d still be thinking about:

Make your job easy so you can focus on the hard stuff:

Farhan kept returning to one key idea: most people make their jobs far harder than they need to be, and it robs them of the ability to do meaningful work. “Your job should be easy so you can focus on the hard stuff,” he said.

He explained that leaders often create unnecessary friction: too many processes, too many one-off decisions, too much just… work. When everything requires effort, the important stuff doesn’t get the mental space it deserves. At Shopify, he pushes teams to automate anything repeated, simplify decisions, and create “good defaults” so people aren’t constantly reinventing the wheel.

The result? Teams spend less time doing work about work — and more time actually building.

A token leaderboard that reinforces the right behaviors:

Most leaders Farhan talks to want to control AI costs. Shopify has taken the opposite stance. “I’m on the side of I need everyone to use AI,” he explained.

Shopify has a token leaderboard that shows how many LLM tokens people are consuming across models like Claude, GPT, Cursor, and others. It’s not there to shame overuse. It exists to normalize heavy AI usage and make it visible. The baseline expectation, as he put it, is that you’re working as if you have a powerful AI assistant beside you. “You don’t mandatorily have to use AI, but we will gauge your impact as if you also have an AI assistant with you,” Farhan said. “It’s like if you and I were playing chess: you don’t have to use a chess computer, but I’m going to. You probably want to use a chess computer.”

Shopify built an internal LLM proxy that exposes all the top models behind a single interface. Engineers (and non-engineers) can choose what they need for the task at hand — from cheaper models to the most capable, high-end ones. The leaderboard then simply shows who is really using the tools available: “The idea is to share how lazy you can be. I want people to say, ‘Look at this thing I built — it only took me five minutes with Claude or GPT.’”

At this particular moment, Farhan would rather over-rotate on enablement now, get everyone “AI-reflexive,” and only later come back to optimize which tools and models people use for which tasks. “I’d rather fight the battle of, ‘Hey, you don’t need the biggest model for that,’ than fight the battle of, ‘Can I convince you to use AI at all?’”

Why Shopify hires 1,000 interns:

While so many companies are cutting back on intern programs and junior hires, Shopify is going the other way. It’s going to hire 1,000 interns this year (300 or so at a time). Farhan drew a direct line from the mobile era to the AI era when talking about early-career talent.

At Twitter, he used to hire a lot of younger engineers onto the mobile team because they had grown up with smartphones; they were mobile native in a way older engineers simply weren’t. The same pattern is now playing out with AI. “ChatGPT is three years old now,” he explained. “The people you hire right out of school had AI for most of their university career.”

That’s a big part of why Shopify decided to hire so many interns. “I hired them because I’m going to learn more from them than they’re going to learn from us,” Farhan said. “They’re more AI-reflexive. They have energy. They’re hungry. They’re intense.”

Interns show up already comfortable with LLMs, cursor-like tools, and AI-assisted workflows. They question legacy processes (“Why do you do it this way?”), push teams to simplify, and model what Farhan calls AI-reflexive behavior — reaching for AI immediately to strip out toil and get to the interesting part of the work.

For complete event notes and a full recording, apply to join Enrich.