While Google, Microsoft, and Meta shed thousands of jobs each at the start of the year, Silicon Valley's developer community was flooded with a different kind of story: engineers saying they couldn't even land an interview anymore because AI had pushed them out of the running. The claim that AI was replacing software engineers circulated as something close to settled fact, backed by a steady drumbeat of statistics showing tech hiring postings down sharply year over year. But when venture capital firm SignalFire analyzed actual hiring data from tens of thousands of companies, the numbers told a different story. Even at the peak of layoff anxiety, engineering's share of new hires actually rose.
When the story the public accepts and the data on the ground diverge this sharply, it says something is off in how we talk about jobs in the AI era. Following that thread doesn't lead back to a story about engineering — it leads to a more basic question: what kind of work are you actually doing?
The hiring increase hiding behind the layoff headlines
According to SignalFire's analysis, tech companies kept hiring engineers steadily even as the AI transition accelerated. Engineering's share of total new hires climbed relative to other job functions. Demand was especially pronounced for developers, machine learning engineers, and infrastructure architects who could directly design and operate AI systems — while roles centered on routine work moved in the opposite direction.
There's a predictable logic behind this pattern. The more companies adopt AI systems, the more people they need to build, maintain, and improve them. It's the same logic that plays out on a factory floor: bring in automated equipment, and demand for the engineers who service that equipment rises too. AI generating code doesn't erase demand for the code — and the people — that build and manage the AI itself. As AI systems grow more complex, the engineers responsible for them become more specialized, not less necessary.
Look at where the layoffs actually concentrated, and a different pattern emerges. Cuts clustered in parts of recruiting and HR, in repetitive marketing operations, in high-volume content production roles, and in entry-level customer support. This looks less like the sweeping narrative of "AI eliminates jobs" and more like AI displacing certain kinds of repetitive work while simultaneously creating new kinds of work elsewhere. The dividing line isn't the name of the job function — it's the nature of the work itself.
Why this data is hard to read as pure good news
Still, it would be a mistake to treat this data as an immediate reason for reassurance.
Start with the limits of SignalFire's sample, which skews heavily toward venture-backed tech companies. Whether engineering roles at mid-sized manufacturers, retailers, or service businesses — sectors that make up a large share of the Korean economy — show the same pattern requires separate verification. There's no guarantee that global Big Tech's hiring trends translate directly to Korea's small and mid-sized business market. Hiring trends at Seoul-based IT firms have little reason to match labor demand at regional manufacturers.
The more important counterargument sits inside the hiring increase itself. The engineers in demand are much closer to engineers who "build and evaluate" AI than engineers who simply "use" it. As AI code-generation tools spread, junior developers who once handled simple, repetitive coding are seeing that part of their role shrink, even as demand concentrates around engineers who can design and oversee entire AI systems — a quiet polarization happening within the same job title. Engineering surviving as a category is a completely different statement from everyone in that category being safe. Two people can carry the same title and land on opposite sides of this line, depending on what they actually do.
It's also true that this data offers little comfort outside technical roles. Functions like paralegal support, junior content planning, and sales administrative support — roles built around routine work that AI can automate relatively easily — are showing signals of losing ground in the hiring competition across multiple sources. SignalFire's data describes what's happening in engineering; it doesn't underwrite optimism for every job function.
What matters is the layer of the work, not the name of the job
Read through the lens of Korean solo entrepreneurs and freelancers, SignalFire's data points to a single takeaway: what matters isn't the job title itself, but what you actually do inside that job.
Dig into why engineering survived, and it comes down to one common thread. Most of the work done by the engineers who got hired involved breaking down ambiguous problems, verifying AI-generated output, and folding the demands of multiple stakeholders into a single coherent system. It wasn't the role of passing along whatever AI produced — it was the role of evaluating, correcting, and adapting what AI produced to fit the context. The people who survived weren't the ones using the tool called AI. They were the ones judging what the tool produced.
What happens when you carry this observation over to people who aren't engineers — planners, content directors, solo founders? Thinking through the list of things AI still doesn't do well points to a direction. Building trust with a client who has complicated, competing interests. Making a judgment call when there's no data or numbers to lean on. Managing expectations that have quietly formed over a long relationship. Sensing a need the other person hasn't put into words yet. These remain areas where humans still take the lead.
The capabilities that keep coming up in forecasts of the post-2030 workplace point in the same direction. Not the ability to operate a tool, but the ability to judge what that tool produces within context. Not the ability to churn out output efficiently, but the ability to build trust between people. Not the ability to follow a set procedure quickly, but the ability to keep learning fast in unfamiliar situations. These are capabilities that apply whether you're running a café, planning content, or running a small consulting practice — they don't care what industry you're in.
I don't think these capabilities are new. They were strengths held by people who'd been doing good work long before AI arrived. What's changed is that, after AI, those same strengths have become a far sharper point of differentiation than before.
A checklist you can actually run right now
From the standpoint of a solo entrepreneur, this data suggests a few concrete questions worth sitting with.
Start by taking an honest look at how much of what you do is routine work that AI can already perform at a comparable level. If drafting reports, repetitive image editing, standardized customer responses, or routine data organization make up a large share of your time, it's worth thinking now about where to redirect that time. Testing directly, up front, how far AI can already go with that work gives you a real baseline to work from.
The scarcer skill is shifting away from "have you used AI tools" and toward "how sharply can you review what AI produces." The gap is already widening in the market between people who pass along AI-generated text or analysis as-is, and people who can catch its errors and biases and refine it into something better. The ability to review a tool's output is becoming more valuable than the ability to simply operate it.
It's also worth taking stock of how you're building trust through relationships. Long-standing clients, repeat business, referral networks — these are assets that AI can't reproduce quickly. As technology levels the playing field and access to AI tools becomes universal, the weight of these relationships only increases. Among competitors using similar tools, the differentiator stops being the tool and becomes the relationship.
What SignalFire's data shows is the survival of one particular job category — but tracing the reasons behind that survival leads to a broader observation: what matters more than the name of your job is the actual character of the work you do within it. Rather than looking at layoff statistics and concluding "engineers are safe" or "my job is at risk," the more useful starting point for a career strategy is a concrete inventory — sorting out, specifically, which parts of what you do now can be handed to AI, and which parts need to stay with a person.
