Innovation & Tech

Software engineers face new job requirement: talking to people

AI now writes much of the code, so companies want engineers who meet customers, present, and judge output — and junior hiring is collapsing.

By Daniel Okafor

5 min read

Coders have to learn a new skill: Socializing
Coders have to learn a new skill: SocializingThis_is_Engineering / pixabay

What's News

  • About 1.9 million people worked in US software development and QA last year, up from 1.5 million in 2020, per BLS.
  • Communication skills appear in nearly 60% of software engineer job descriptions, up from 47% in early 2020, per CompTIA.
  • Robert Half data shows in-office requirements in 84% of tech postings this fall, up from 63% in early 2024.
  • Anthropic's junior technical hires fell from 27% in March 2025 to 9% in May; engineers-to-PM ratio hit 17 to one.

By Daniel Okafor

Nearly 60% of open software engineer roles now list communication skills in job descriptions, up from 47% in early 2020, according to CompTIA. That single statistic explains what is happening to one of the largest technical professions in America.

Michael Irwin, a principal software engineer at Docker, spent recent weeks on a woodworking project. He handmade a wooden base for a gaming controller, wired red, blue, and green buttons to it, and carried it to a conference last week. The game behind it was coded by AI. Irwin built the prop to pull people to his booth and start conversations about AI agents. "It's been a dream of mine for a long time, and it's just now more accessible than it's ever been," he says.

The booth gimmick marks the shift. AI now generates much of the code, so engineers spend their time on judgment, design, and conversations. "The micro level is mostly able to be passed off," Irwin says. "It takes a different mindset."

What changed in the engineering job?

Three developments reshaped the role. Vibe coding changed software engineering last year. Microsoft and Google said AI had written large swaths of their latest code, and engineers began reviewing more AI-generated lines than they typed. Amazon Web Services recently used agents to generate 100% of the code for a new emulator tool. Andrej Karpathy, the father of vibe coding, called 2026 the year of "agentic engineering."

Second, non-engineers started writing code. A fifth of OpenAI's Codex users were non-developers as of June. Third, bosses now expect developers to manage teams of agents, explain technical work in plain terms, and act as architects and designers rather than code producers.

"This new way of working, it might not suit the typical engineer personality," says Sam Dhar, an engineer at Galileo, an AI startup Cisco acquired earlier this year. He notes the pressure is real: developers who fail to make the shift on a compressed timeline risk layoffs, replaced by candidates who can charm an interview rather than ace a take-home coding test.

Is the profession shrinking?

No — so far. About 1.9 million people worked in software development, testing, and quality assurance in the US as of last year, up from about 1.5 million in 2020, according to the Bureau of Labor Statistics. The bureau projects 10% growth over the next decade.

The work itself has changed. Dylan Wight, founding software engineer at Flagler Health, an AI-native operating system for musculoskeletal care, spends his days reviewing code issues raised by agents and sorting practical problems from oversensitive flags. He also holds multiple meetings weekly with healthcare users. "If people wanted to just think about code, I think it would get harder, because that part is mostly done," he says.

Paul Elliott, vice president of product and engineering at Chromatic, says the job demands more business context: "You have to really understand a lot more about why we're doing whatever it is we're doing, and ask a lot of the right questions." He adds: "It's definitely a trainable skill."

How far does the engineer–manager wall break down?

Consider Anthropic's internal data. A workforce report the company released last month shows technical workers spend 16% more time in meetings than a year ago, while the ratio of engineers to product managers has jumped to 17 to one. AI companies increasingly title these hybrid workers "member of technical staff." Forward-deployed engineers, who work inside client systems, have grown popular in the AI era.

"Engineers are definitely taking on more product level autonomy, but product managers are also taking on more technical work," says Frank Fusco, CEO of Silicon Society. "I think the definitions of those jobs are changing."

At Amazon Web Services, Marc Brooker, vice president and distinguished engineer, says his engineers now spend more time with customers and prospects. Training happens informally, often by shadowing calls with more public-facing colleagues. The goal, Brooker says, is that "engineers can go back to their teams and their organizations and turn what they learned into really great product decisions, really great technical decisions."

Why are offices back in demand?

Michelle Volberg, CEO of recruiting firm Twill, says about half of the roles her company filled last year were remote. Now roughly 95% require hybrid or full-time office work. Robert Half data tracks the same move: 63% of tech job postings had in-office requirements in early 2024, versus 84% this fall. Remote tech postings fell from 10% in early 2023 to 3%. "A lot of these managers are bringing engineering on site now to have this collaborative approach," says Kareem Osman, tech industry expert at Robert Half.

Some hiring managers now screen for public presence. Volberg searches for developers active on YouTube and LinkedIn who already post about what they build. Chuck Hardy, director of product and engineering at Akeneo, runs role-playing sessions with candidates before panel interviews. "We're not there to test you or to trick you," he says.

What happens to junior engineers?

Here the numbers turn grim. Junior technical hires at Anthropic fell from 27% of the total in March 2025 to 9% this May. About 7 to 8% of recent computer engineering and computer science graduates are unemployed, per the Federal Reserve Bank of New York — on par with performing arts majors. The vanishing training ground could produce not just future worker shortages but fewer people practiced enough to spot flaws in AI-generated code.

Dhar frames the broader stakes plainly: "It's only happening to engineering first. It's potentially going to manifest in a way where all work is done in some respects with AI, where the accountability still resides with the humans who are reviewing and fixing and asking the right questions."

Source: feeds.businessinsider.com

Share this article:

More from Daniel Okafor

Daniel Okafor

Show full bio

Correspondent covering money, funding and market shifts around entrepreneurship, startups and venture funding at Business Bearings.

2,189 articles

Related articles

« Previous articleNext article »