Why engineering jobs are the AI era's most resilient
New SignalFire data says engineers make up a bigger share of new hires, not fewer. Here is what that signal means for a Sri Lankan dev building a career on a tight budget.
The headline everyone repeated for two years was that AI would kill engineering jobs first. New hiring data points the other way: engineers are making up a larger share of new hires, not a smaller one. That is the finding TechCrunch reported from SignalFire, and it is worth more than a doom-scroll because it changes what a Sri Lankan developer should actually do this month.
I want to read the signal honestly, strip the hype, and turn it into a plan you can act on whether you are a UCSC student, a Colombo agency dev, or a one-person freelance shop. Source: TechCrunch — "AI was supposed to kill engineering jobs, but new data suggests they're the most resilient".
🔍 The narrative said one thing, the data says another
The layoff stories and the hiring numbers are not measuring the same thing. Layoffs are loud and easy to write about. Net hiring is quiet and harder to track, which is why a dataset like SignalFire's matters: it looks at who companies are actually bringing in, not who they are cutting.
Key takeaway: "AI is in the layoff headlines" and "engineers are a growing share of new hires" can both be true at once. The first is a story; the second is a signal. Plan around the signal.
Here is the gap, plainly:
| What you keep hearing | What the hiring data suggests |
|---|---|
| AI replaces engineers first | Engineers are a bigger slice of new hires |
| Coding is being automated away | Companies still compete to hire people who can build |
| Junior roles are vanishing | The bar moved, but the door is not shut |
If you have been holding off on learning to build because "AI will do it anyway," that assumption is the most expensive thing you own right now.
⚡ Why engineers held up better than expected
AI is good at producing code. It is not good at owning a system. Those are different jobs, and the second one is what companies pay for. A model can draft a function in seconds, but someone still has to decide what to build, judge whether the output is correct, wire it into a messy real codebase, and carry the pager when it breaks at 2am.
The resilient skills are the ones that sit around the code, not the typing of it:
- Problem framing — turning a vague business need into a spec a machine (or a junior) can execute.
- Verification — knowing when generated code is subtly wrong, which is harder than writing it correctly the first time.
- Integration — making new code survive contact with an existing system, its quirks, and its data.
- Judgment under constraint — picking the boring, cheap, maintainable option over the clever one.
AI made writing the first draft cheap. It did not make being responsible for the result cheap. That responsibility is the job.
This is also why the "AI replaces juniors" fear is overstated but not zero. The floor rose. A junior who only copies tutorials is now competing with a free model that copies tutorials faster. A junior who can read a model's output critically and ship a fix is more valuable than that same junior was three years ago.
🌐 What this means specifically for Sri Lankan developers
For a developer here, this data is unusually good news, and not for the obvious reason. The cost of becoming dangerous at this work has collapsed. The expensive parts of a tech education, a mentor who reviews your code and a fast feedback loop, are now partly available for the price of an internet connection.
| Career input | Cost a few years ago | Cost now |
|---|---|---|
| Code review / mentor feedback | Senior dev's time | An AI pair you can question 24/7 |
| Learning a new stack | Bootcamp or course fees | Free docs + a model that explains them |
| Building a portfolio | Weeks per project | Days, if you direct the tools well |
The leverage point for an SL dev competing for remote work is no longer "can you write the code." Plenty of people and tools can. It is "can you be trusted to own an outcome from a different timezone." That is a verification-and-communication skill, and it is exactly the resilient skill the hiring data rewards.
A few concrete moves:
- Build in public. A GitHub profile with three finished, deployed things beats a CV full of course certificates.
- Charge in USD where you can. Remote demand for builders is the whole point of this data. When you do land paid work, know your real take-home after platform fees with our Freelancer USD-LKR earnings calculator.
- Use the free dev tools you already have. Cleaning data, decoding tokens, converting types, our free developer tools cover a lot of the small friction so you spend your time on the parts a machine cannot do for you.
💡 The trap: "AI does it, so I don't need to learn it"
The most dangerous reading of this story is relief. "Engineering is safe, good, I can relax." That is not what the data says. It says the people who build and own software are doing well. It says nothing kind about people who avoided learning because they assumed the field was dying.
Bottom line: The hiring data does not protect engineers as a job title. It rewards the ability to direct, verify, and ship. Become that person and the trend is your tailwind; skip it and the same trend passes you by.
If you are a student, the practical version is blunt: let the AI explain the concept, then close the tab and rebuild it from memory. The understanding is the asset. The generated code is disposable.
🚀 What this means for you
You do not need to predict the AI jobs market. You need to be on the right side of one specific split: between people who produce code and people who own outcomes.
- If you are a student: keep learning to build, hard. The floor rose, so clear it. Ship small real projects, not tutorials.
- If you are a junior dev: practise verification. Get fast at spotting where AI output is wrong. That is now a senior-level skill you can start building today.
- If you are a freelancer or small-team builder: lean into ownership and timezone-independent trust. That is the moat AI does not cross, and it is the exact thing remote clients pay a premium for.
The story you were told was that AI comes for the builders first. The data so far says builders are holding the line, as long as they keep building. That is a fair deal. Take it.