It’s a tricky job marketplace for software program engineering hopefuls. Tens of hundreds of jobs cuts throughout the trade have elevated competitors for open spots. The rise of AI has spurred fears of dishonest throughout interviews, and firm priorities are altering because the tech evolves nearly every day.
However hiring managers have a much bigger concern: Now that AI can write code, how can you determine who – and even what skillset – makes a superb software program engineer?
The interview course of hasn’t stored up with the best way AI has modified programmers’ every day obligations, profession consultants and software program engineers informed CNN. That’s made the hiring course of more difficult for each job seekers and hiring managers.
“I’d say AI has hit engineering interviewing like an atomic bomb,” mentioned Stefan Mai, a former engineer for Meta and Amazon and cofounder of tech interview teaching service Hey Interview.
Software program engineering has been among the many first industries to be noticeably impacted by AI. A report from Google’s analysis division final yr discovered that 90% of tech workers use AI for duties like writing and modifying code, up 14% from the earlier yr. The sphere has been intently watched as a bellwether as AI adoption expands.
AI can now assist software program engineers write code and documentation; analyze knowledge; be taught coding ideas and troubleshoot, amongst different issues. That lets tech corporations transfer far more shortly, some executives say.
An OpenAI engineer used AI to implement a system change that may have taken his crew every week in any other case, firm president Greg Brockman recently said throughout a Sequioa Capital discuss. Many inside apps at Google are being “principally” written with the corporate’s Antigravity AI coding software, Varun Mohan, a director at Google DeepMind, informed CNN earlier this month.
Boris Cherny, Anthropic’s head of Claude Code, wrote on X in December that “100%” of his contributions to the product over the previous 30 days have been written by Claude Code.
Cherny believes AI is shifting the position of a software program engineer to give attention to high-level decision-making reasonably than writing code. The title “software program engineer” is likely to be changed with a reputation like “builder” that higher captures the place, he beforehand informed CNN.
AI isn’t meant to exchange engineers, Google’s Mohan informed CNN.
“We predict builders ought to spend most of their time attempting to determine what they need to construct,” he mentioned. “That’s the entire query.”
Madhu Kurup, vice chairman of engineering at Certainly, in contrast AI in software program engineering to Google Maps’ position in journey. Google Maps can inform an individual which freeway exit to make use of, flag visitors situations and discover espresso outlets on a driver’s route, however it doesn’t choose the vacation spot or resolve what time to depart.
Nevertheless, that Google report final yr indicated that 46% of tech employees solely “considerably” belief the standard of AI-generated code, and 31% mentioned AI solely “barely” improved code.
And layoffs all through 2025 and 2026 counsel AI is affecting payrolls. AI was the highest purpose cited by corporations for job cuts in April for the second month in a row, the manager outplacement agency Challenger, Gray & Christmas mentioned this month.
Corporations for years have evaluated potential candidates with regimented tests that some say have felt like taking the SATs, however for coding. They don’t measure how employees delegate duties to an agent and use AI to navigate points, brainstorm or work extra effectively.
Now, some engineers really feel these exams not replicate “what their job will really seem like,” mentioned Jordan Leonard, cofounder and chief working officer of Leopard.FYI, a tech hiring community for girls and genderqueer engineers.
In late April, software program developer David Barajas mentioned he had been on about 5 – 6 job interviews after being approached by recruiters over the previous six to eight months. None of them requested to see how he incorporates AI coding instruments like Cursor into his work.
“The very first thing they are saying is, you’re not speculated to be utilizing any AI instruments, no AI help, nothing that can assist you clear up this drawback,” he mentioned.
Sujata Sridharan, who most lately labored on the fintech agency Bolt and has spent roughly a decade as a software program engineer, additionally mentioned many of the corporations she’s lately interviewed with are utilizing the identical conventional exams that target understanding code reasonably than working with AI. “There may be that hole, it’s simply grown wider (due to AI),” she mentioned.
Some corporations initially prohibited candidates from utilizing AI throughout exams to stop dishonest. Barajas mentioned he’s even been requested to share his desktop throughout interviews to show he’s not utilizing AI. Whereas such issues are nothing new, worries about dishonest have “actually ramped as much as an uncommon stage” due to AI, in keeping with Mai.
And AI’s fast development means position necessities are continuously shifting.
One firm Leopard.FYI works with mentioned the programming language Ruby on Rails was required for an open place, mentioned Leonard. Three weeks later, the corporate modified its thoughts as a result of AI can simply translate different languages to Ruby on Rails.
“It looks like a shifting goal on actually a weekly (or) month-to-month foundation,” she mentioned.
Employers are beginning to focus extra on questions that present how candidates assume by points and weigh trade-offs reasonably than uncooked coding capabilities. These subjects historically come up in senior-level interviews however are actually changing into extra frequent throughout the board, in keeping with Mai.
Some startups have experimented with bringing in candidates to work on web site for a half a day. And it’s changing into more and more frequent for managers to permit candidates to make use of AI through the testing course of, in keeping with Leonard.
However even these modifications don’t totally seize how the job will get finished today. Sridharan, for instance, often collaborates with AI to work by points, however her expertise utilizing it in technical exams often entails utilizing it as a substitute for hands-on coding.
It’s nonetheless an “unsolved drawback,” in keeping with Mai.
“It’s sort of surprising what candidates are going to run into,” he mentioned.