Why Companies Are No Longer Hiring Junior Software Engineers

Freshly graduated software engineers are facing a harsh job market as artificial inteligence tools reshape how software is built and push companies to favor experienced programmers over entry-level hires.

Gone are the days when developers wrote code line by line. Currently, most of them generate large portions of it simply by instructing AI systems. “My colleagues say they don’t remember the last time they wrote code without AI,” Nikita Kuznetsov, founder of GameAP.com, told Nearshore Americas.

The shift is nearly universal. Laura Tacho, former chief technology officer at DX, a firm that studies how engineering teams work, told a summit in February that AI already generates between 25% and 40% of the code used at many companies.

The math is not lost on talent managers. “Companies are not hiring five junior engineers when one senior engineer with the right AI tools can outproduce all five,” said Elmer Morales, CEO of Koder.com in an interview with Nearshore Americas. “Junior engineers who skip the fundamentals and go straight to AI-generated output are building on a foundation they cannot fix when it breaks. That is a real and growing problem.”

The result is a sharp contraction in entry-level hiring. The number of graduates brought on by U.S. tech firms has dropped by nearly 50% compared to pre-pandemic levels, according to a study by venture capital firm SignalFire. Oracle has gone further: The company has withdrawn campus job offers already made to students at the prestigious Indian institutes of technology (IITs), the places that produced executives like Google CEO Sundar Pichai.

Engineers or supervisors?

As AI absorbs the repetitive and mechanical work of coding, developers are increasingly functioning less like programmers and more like overseers. “The best engineers today are not writing every line. They are directing, reviewing, and making architectural decisions. The job has changed,” Morales said.

The analogy that exemplifies better the new reality is construction: a site supervisor who guides the crew rather than laying the bricks. “Right now I’m mostly managing what the AI does and reviewing the code it generates. During that review, I also rely on AI’s help,” Kuznetsov said.

“My workday now is more like ‘design this system, make sure the AI got it right’ than ‘code this function,'” said Arpit Sabherwal, a software engineer at fintech Plaid, in written comments to Nearshore Americas.

None of this means AI has made software engineering accessible to all. Companies still need engineers precisely because it cannot fully grasp the specific needs and constraints of a business. “AI can be used to expedite repetitive tasks such as writing boilerplate code, composing test cases, explaining unfamiliar code, and recommending small changes,” said Louis Leung, co-founder of inFlow Inventory. “However it is not a replacement for the ability to build clean systems, understand intricate business logic.”

Leung, who studied computer science at the University of Waterloo, argues that AI can identify patterns in past data but still struggles with the deeper reasoning behind how systems work. Humans remain better at solving complex problems, building systems that adapt as business needs evolve, and designing security setups for specific situations.

“You can’t just tell the AI to ‘do this well’ or hand it a technical spec. You have to break the task into parts and give the AI concrete instructions,” Kuznetsov added.

That is precisely where new graduates are stuck. Only experienced developers know how to give AI those concrete instructions, because they learned to build software before the arrival of AI. Fresh graduates never had that chance. And without an entry-level job to develop it, they may not get one.