In the AI Era, Shopify Is Investing in Junior Engineers—Not Cutting Them
We spoke with Farhan Thawar, VP and Head of Engineering at Shopify, about why the company 10x’d its internship program, what AI-native hiring really looks like, and the three-part framework reshaping how technical talent gets evaluated.
The conventional wisdom goes something like this: AI is coming for developers, and the first casualties will be the junior engineers — the interns, the new grads, the people who used to spend their days writing boilerplate and fixing typos in documentation. Why hire a cohort of entry-level engineers when a model can do their work in seconds?
Shopify isn’t buying it.
Last year, Shopify didn’t quietly trim its internship program. It exploded it — growing from roughly 100 interns a year to over 1,000. This year, they’ll continue the program with plans to hire 1,000 interns. We spoke with Thawar to understand why, what it tells us about the future of technical hiring, and what early-career engineers can do right now to stand out.
Why +10x the internship program — and why now?
Shopify has always run an internship program. The philosophy hasn’t changed: bring in next-generation talent, keep the company thinking with fresh eyes, and make sure there are always people in the building willing to ask “why do we do it this way?”
The wrong answer is ‘because we’ve always done it this way.’ The right answer is ‘good question — let me explain, or actually, I never thought about whether there was another way.
5 stars
But the scale shift is new, and AI is part of the reason. Thawar draws a direct parallel to the mobile era: when smartphones took over, Shopify deliberately hired people who had grown up on mobile — people for whom the phone-first mental model was instinctive, not learned.
“These folks coming out of schools now are growing up with AI,” Thawar says. “They are in school with AI the entire way through. In many ways, they’re AI native. We wanted to bring those types of people in to reimagine what it looks like to build knowing they’ve grown up with AI.”
There’s also a cultural dimension. Interns bring intensity, curiosity, and energy — and a cohort of 350 spreads that energy far more effectively than 25 ever could. “Twenty-five engineers will have a harder time impacting us than 350,” Thawar notes. The interns are distributed across teams, not siloed into a single program track.
And practically: Shopify converts a significant number of interns into full-time hires. The four-month internship functions, in Thawar’s words, as a “two-way interview” — the company evaluates how candidates think about problem-solving, how they wield AI tools, and how they operate under real conditions. The interns get a genuine sense of Shopify’s engineering culture and the resources available to them.
There’s one more benefit Thawar didn’t expect to become a talking point: the energy in the office. Shopify is a remote-first company that gathers intentionally rather than requiring daily attendance — but interns come in every day. “You’re at lunch and you see all these amazing interns around and you can pick their brain,” Thawar says. Leadership teams flying in for company gatherings now overlap with a room full of curious, energetic early-career engineers. That collision has shaped Shopify’s in-person culture in ways the team didn’t fully anticipate.
What does ‘AI-native engineer’ actually mean at Shopify?
Some companies have responded to the AI moment by rebranding their engineers. “AI engineer” is showing up on job descriptions everywhere. Shopify doesn’t do this — and the reason is telling.
“For us, it’s implicit,” Thawar explains. “If I call some people AI engineers, then other people will be like, ‘does that mean the finance person is not an AI finance person?’ Everyone should be using this tool.”
The expectation at Shopify is that every engineer understands how to build software — and that they also understand when to use AI, when to trust it, when it’ll give them a novel approach, and how to validate its output. Critically, whoever submits a pull request owns that code.
You can use AI tools, but you still put your name on the PR. A human reviews it and puts their name on it too. You have the responsibility of the code it generates once you submit it.
5 stars
This isn’t about limiting AI use — it’s about not outsourcing judgment. Shipping fast matters. So does shipping right.
What can engineers do now that they couldn’t do two years ago?
Thawar gets enthusiastic about this one. Three things stand out in his answer.
First: managers can code again. “The ramp-up is so much lower now. I might have 15 minutes between meetings and I can pull up Claude Code or Cursor and quickly build a prototype, or ask it questions about the existing codebase.” In the past, meaningful code contributions required blocked-off time — a full day, a sprint week, a dedicated “no meetings” stretch. AI pair programming has shrunk that activation energy to nearly zero, enabling managers, directors, VPs — and at Shopify, the CEO — to contribute working code regularly.
Second: the work you always wanted to do but kept putting off. Writing unit tests. Refactoring gnarly legacy code. Wholesale rewrites of the modules that everyone quietly dreaded. “You can have a conversation with your agent: let’s reimagine this domain. How would you approach it? Then say: let’s code that — and actually look at the prototype.” That was never accessible before because the activation energy was too high. Now it is.
Third: ambition has no excuse. “In the past it would’ve taken me weeks and weeks to learn a new tool chain. With AI, you can just go after it.” Nine out of ten prototypes might go nowhere — but one might hit, and now the barrier to finding out is almost zero. Engineers and non-engineers alike can pursue the feature they always thought was missing.
Shopify’s hack days have transformed as a result. “It’s more likely now that you’re going to actually build something,” Thawar notes. “People used to spend hack days learning how to use data at Shopify or how to use a tool. Now it’s: I’m going to learn it and build something I can at least demo.”
How Shopify evaluates technical talent in the AI era
Here’s where Thawar gets refreshingly candid: “I have an answer I don’t love. The honest answer is: we don’t know. No one has really figured out yet what it means to evaluate the next generation of technical talent with these tools.”
What they do have is a framework — one Thawar borrowed from the University of Waterloo — that’s starting to shape how Shopify thinks about interviews. It has three modes:
No AI allowed. Can you write code by hand? Do you understand what’s happening at the layer below where you’re working? This is how software was done just a few years ago, and Shopify still wants to know you can operate there.
AI optional. Do you know when to use it? Can you make a judgment call about whether to pull in the AI or push further on your own thinking first? This is about discernment, not just capability.
AI mandatory. The project is too big for the time available. The candidate has to wield the tool effectively — scope, prompt, validate, ship. Imagine being asked to build a full Twitter client in an hour. With AI, you can get surprisingly far. Without it, you can’t.
‘This move feels good’ — some chess players can’t articulate every tactic, but they consistently make the right call. I think something like that exists in software now too.
5 stars
The framework is appealing but imperfect, and Thawar is the first to admit it. There may be a new category of developer — people who don’t deeply read the underlying code but can build extraordinary things through an agentic loop they’ve mastered. Whether that style of work should pass a traditional evaluation is an open question. So is the more fundamental one: for candidates who score well on these techniques, do they actually perform well on the job afterward? “I don’t think anyone has cracked that in any part of the software industry yet.”
What early-career engineers should do right now
Thawar’s advice for students and early-career candidates hasn’t changed in 30 years of working in the industry — but AI has removed the last remaining excuses for not following it.
Build something. You used to have a little bit of an excuse — you had to read the APIs, learn mobile development, figure out desktop. Now with AI, you have no excuse.
5 stars
Build tools for yourself. Build tools for people you know — Thawar built an options-trading algorithm for his 84-year-old father (who declined to use it, preferring paper and pen, but that’s beside the point). Put it on GitHub. See if other people need it. Contribute to open source — find issues, try to fix them. Build against Shopify’s API if you want to work at Shopify.
“There is no shortage of work to do in the world. Start working on it. You don’t need a job to do that.”
The portfolio matters more than the resume. As Thawar’s former CTO observed: a resume tells you what someone did, but never why. Working software tells you much more — and for intern evaluations, the ultimate signal is real impact. An intern who deleted six lines of code and saved Shopify $600,000 in infrastructure costs didn’t show up in any activity metric week to week. But the impact was undeniable.
What this means for technical hiring broadly
The dominant narrative around AI and developer jobs runs something like this: AI will automate the routine work, and the routine work is what entry-level engineers do. Ergo, fewer entry-level engineers.
Shopify’s internship expansion represents a direct challenge to that story. Thawar’s bet isn’t that AI makes junior developers redundant — it’s that AI makes developers who grew up with AI more valuable than ever. The people who need the least convincing to wield these tools, who find the mental model natural rather than foreign, who ask “why are we doing it this way?” with genuine curiosity rather than learned skepticism.
That’s a different kind of talent signal. And if the University of Waterloo’s three-part framework — no AI, AI optional, AI mandatory — becomes an industry standard for evaluating technical candidates, platforms like CoderPad are positioned to surface exactly that signal: not just whether a candidate can code, but whether they know when to reach for the tool, when to set it down, and what to do when it’s gone.
The developers who figure that out first won’t be the last to be hired. They’ll be the first.