.jpg)
How AI Changed the Coding Interview
Software engineering used to mean memorizing syntax and grinding through algorithm puzzles under a watchful eye. That world is gone. Today's hiring managers assume candidates will use AI tools during interviews — the question isn't whether you can code without help, but whether you understand what the AI is doing well enough to catch its mistakes.
Le'ale Addison, a 22-year-old computer science grad who's interned at Amazon, KPMG, and Smurfit Westrock, watched this shift happen in real time. Just two years ago, she had to share her screen to prove she wasn't cheating. By the time she graduated, using AI or Google mid-interview was standard practice — and interviewers had started asking pointed questions about her familiarity with machine learning and how she used AI in her actual workflow.
Recruiting has moved to X and GitHub
Companies aren't just changing what they ask — they're changing where they look. Executives at AI startups like Cognition, Base44, and Replit say they now scout talent by browsing X and GitHub instead of résumés. Replit's chief people officer described X as becoming a primary recruiting channel, with employees flagging promising engineers they've met at conferences or connected with online. Cognition's head of people said she's personally traveled the world to recruit standout engineers, going as far as driving a candidate to the airport just to get more time to make her pitch.
The technical interview isn't dead; it's different
LeetCode-style puzzle tests, once the industry standard, are now just a first filter — if they're used at all. Xavier Contreras, a data engineering lead at a hedge fund in New York, said his most recent job search looked nothing like the ones from five or six years back. Instead of pure coding challenges, he was asked to explain his architectural reasoning and defend design decisions. Take-home projects that once took a month now get compressed to three days, since AI handles the grunt work.
Companies say the real skills they're testing for now are systems thinking, judgment, and the ability to operate agentic AI tools effectively. Cisco, for instance, has shifted from coding challenges toward project-based exercises that put candidates inside AI-assisted development environments to see how they actually work.
Culture fit and "data unicorns"
Technical rigor hasn't disappeared — if anything, the bar for elite roles at places like OpenAI and Anthropic keeps climbing, according to career coach Sundeep Teki. But companies are weighing culture fit more heavily too; Anthropic reportedly runs a dedicated non-technical interview focused purely on mission alignment.
Some companies skip theory entirely and just have candidates start working. Firms like Lovable, Cursor, and Kilo run in-person trials where candidates ship real product within a day or two — sometimes candidates take vacation time from their current job just to attend.
The bigger picture: as AI automates much of what engineers used to do by hand, the roles themselves are merging. Contreras said software engineering, data analytics, and data science have effectively folded into one job description, with companies looking to hire a single "data unicorn" who can do the work of three specialists.