AI tech hiring shifts from bootcamps to reskilling

A decade after governments, bootcamps and employers told workers to learn code as a reliable route into tech, artificial intelligence is starting to unwind that promise. The clearest signal is not a single mass layoff announcement. It is the collapse of older entry paths at the same time that hiring tilts harder toward engineers and companies ask existing staff to absorb more AI work.

ZDNet captured that shift in its report on the closure of Kentucky training programme Code:You, after its leaders found that demand for training held up while entry-level jobs did not. That same week, Business Insider’s analysis of SignalFire data argued that software engineers now account for 55 per cent of hiring at major tech companies, up from 46 per cent in 2019, even as design, product and marketing roles lost share.

For Australia, that combination matters because it points to a narrower, more expensive labour-market shift than the old “AI will take every job” debate suggests. The more immediate change is that employers still need technical people, but they want fewer narrow specialists, more AI-fluent generalists and a workforce they can retrain in-house. That is a different problem. It lands on enterprise budgets, university pipelines and mid-career workers all at once.

The old “learn to code” pipeline is breaking at the bottom

The first view comes from inside the training market: the people who built programmes around the last decade’s shortage of developers are now seeing the market change before the headline employment data fully does. In ZDNet’s report on Code:You, the training demand never really disappeared. The jobs did.

“There was no shortage of interest from job seekers looking for training; it was just the jobs that were available to entry-level people in the field that seemed to have dried up.”
Brian Luerman, Code:You, via ZDNet

What makes that distinction important is the way the hiring market changed underneath it. During the cheap-money years, the sector could tolerate broad hiring funnels, long junior ramps and specialist teams whose work sat one step removed from revenue. In the AI cycle, companies appear to be buying productivity first and sorting out headcount second. Business Insider’s reporting on software-engineering interviews says employers now want more than raw coding speed; they are screening for judgment, systems thinking and the ability to use AI tools without being carried by them.

That leaves the cost question. The bootcamp-era answer was mostly the worker. The new answer is increasingly the employer, whether they like it or not. ABC’s reporting on an Australian government review of AI and jobs cited forecasts showing 77 per cent of employers plan to upskill workers, even while 41 per cent expect AI-driven workforce cuts. Those numbers are easy to read as contradiction. They are really a description of the same transition: some roles thin out, while the people kept on staff have to do more, faster, with different tools.

“It’s like a real-time experiment to say, as we integrate AI into the work, what happens to jobs, skills, team size, career trajectories?”
Julie Bedard, via ZDNet

AI is changing the org chart more than it is ending tech work

The analyst’s counterpoint is unavoidable. If this were only a story about disappearing jobs, the data would look simpler. Instead, the pattern is one of reshuffling. Business Insider’s review of SignalFire’s figures says software engineering’s share of hiring at major tech groups rose to 55 per cent this year from 46 per cent in 2019. Over the same stretch, design hiring fell 48 per cent, product management 39 per cent and marketing 36 per cent, according to the same dataset.

Engineering is not suddenly swallowing every function. But it is absorbing work that used to sit across several adjacent ones. Product managers are expected to prototype. Designers are expected to work closer to tooling. Marketers are being pushed toward automation stacks and performance analytics. Smaller teams can ship more. But only if the people in them can cross boundaries.

ZDNet’s follow-up on where big-tech hopefuls may still find openings points in the same direction. Startup hiring has held up better in some pockets, but roles outside software engineering remain tougher. In other words, AI is not producing a flat labour-market contraction. It is producing a steeper premium on the parts of a tech team that can directly shape, supervise or extend automated systems.

The founder view sharpens that point. GeekWire’s roundup on Microsoft’s reset and smaller AI-era startup teams and TechCrunch’s profile of an AI office-suite startup built with about 45 staff, including 18 engineers both point to the same operating model: leaner companies, heavier technical cores, less patience for layers of coordination. That model still hires. It just does not hire like the previous one.

For early-career workers, that is the most painful part of the shift. Business Insider reported that tech staff are spending nights and weekends learning AI tools because they think they cannot afford not to. That is how the shift reaches individual workers. The ladder into tech is still there, but more of the rungs are moving.

Australia’s window to respond is smaller than it looks

Australia’s own labour data still gives policymakers some breathing room. ABC reported on 8 July that a federal government report did not find broad AI-driven job losses across the economy. At the same time, that same reporting cited a 25 per cent rise in Australian software-development roles since November 2022. Read together, those numbers suggest the local market is not collapsing. It is becoming more selective.

“When the numbers are showing significant impact of AI on jobs, it’s already too late to respond. We actually have to be proactive.”
Toby Walsh, UNSW Sydney, via ABC

From Walsh’s perspective, the warning is as much about timing as technology. Waiting for a clean job-loss signal misses the practical shift already under way inside teams: more automation in day-to-day work, flatter structures, fewer purely junior tasks and a stronger bias toward employees who can pair domain knowledge with AI tooling. That is why the bootcamp story and the org-chart story are the same story. One describes the shrinking intake valve. The other describes the new shape of the organisation on the other side.

Australian employers therefore face a more operational problem than a philosophical one. Buying finished talent alone will get more expensive, especially in AI, cloud and software roles. The cheaper option may be structured internal retraining, clearer apprenticeship paths and more deliberate use of junior staff on work that AI can accelerate but not fully own. For workers, the uncomfortable reality is that “learn to code” has become “learn to work with code, models and systems”. Broader, fuzzier, and harder to package into a six-month promise.

Nor should this moment be read as a simple victory for engineering over every other discipline. Companies that hollow out product judgment, design sense or customer understanding can move faster for a while and still build the wrong thing. But the hiring data does show where managers think the bottleneck is right now. It is not at the level of headcount in the abstract. It is at the point where AI output still needs technical oversight, business context and someone willing to keep learning after the formal training ends.

The old coding-boom pitch was that the market would meet workers halfway if they acquired the right skill. AI is making that bargain less generous. Employers still need talent. They are just redefining what counts as ready, and quietly moving the cost of getting there back onto themselves.