This matches my experience. About 90% of the coding tasks I would have assigned to a junior in years past, I can get done for a fraction of the price at a similar quality and in 1/3rd the time by a tool like Claude Code. The only reason I have to hire a junior would be to develop them into a senior over 3-5 years, and companies are far too shortsided to think about that training pipeline.
A good senior engineer still does things the tools can't- in particular, non-programming tasks like planning, persuading, and accountability. I fear that there will be a crisis in these skills as the current seniors retire/change industries in the coming years.
If you've as an industry blocked the pipeline saying we don't need these people, time passes and those people who had to invest to get to that point abandon the bad investments, and you get no new people coming in. People leave such industry because of burnout, and aging/retirement/death. The dynamics created are one of a deflationary sieve where the front-loaded cost savings becomes a back-ended cost sink. Tthe demand and related cost of the professionals is recognized at a point where there is great need but they can't be found at any cost. In system's engineering we call this Hysteresis whipsaws, Mises might call this the first early stages of the Economic Calculation Problem.
Well and, without special arrangements, once trained for 3-5 years they'd now be more valuable and as easily scooped up by another company - if companies are short-sighted about the training pipeline, so too are employees short-sighted about how long they intend to remain somewhere
Question I still don't have an answer to is what that career trajectory looks like. I was at a event last week where someone from a popular startup, Clay, described a position (a "GTM engineer") as very intentionally not strategic, but about boots on the ground execution. I came away from the event wondering was "ok, how long until you automate away this line of work?"
I know it's common advice for students and newly grads to do this, but in my experience, employers do not care about personal software projects or open-source contributions unless the work is aligned with their product. That, or you built something that is easily lucrative. Otherwise, they do not care, they do not care, they do not care.
If your goal is personal enrichment, by all means, but don't kill yourself on a personal project with the intention of impressing an employer.
Projects are just about the only thing you can add to your resume to show competence. Everyone has a degree, and no one has work experience.
Not only did my projects show that I have completed semi-relevant work in something considered relatively complex, which presumably got me past resume screenings (confirmed by my hiring manager), my projects also gave a prime talking point in interviews that let me showcase my domain knowledge and way that I work. (Consider cliche interview questions like “what is the biggest challenge you’ve faced”, and how they relate to your projects)
This is especially beneficial if you’re being interviewed by other engineers, and you can geek out over a project. Being human and enjoyable, while demonstrating technical competence, is a great interview winner.
Ofc I have limited experience, but small samples can add up if other evidence corroborates with it.
The ROI is simply bad compared to grinding LeetCode/NeetCode/Cracking The Coding Interview and learning how to game the interview process. This should be every new grad's priority if they are interested in employment. It's even worse than just having gone to a highly reputable university, ideally with a pipeline to FAANG companies.
I have no interest in looking at their resume, the first thing I do is look to see if they have a Github and what they've done with it
My green bar on Github and open source contributions have gotten me everything in life. Money, jobs, contracts, community support, etc.
All the sympathies to these people.
However people keep rejoicing their use of Claude without looking into the future.
Those cool agents that close tickets on their own, eventually will have an error rate lower than humans.
How many architects does a project need?
So I get it, there's not much left...but expecting six figures for your first job is crazy.
Even within the article's anecdotes:
> Ms. Mishra, the Purdue graduate, did not get the burrito-making gig at Chipotle. But her side hustle as a beauty influencer on TikTok, she said, helped her realize that she was more enthusiastic about tech marketing and sales than software engineering. [emphasis mine]
Sounds like she was headed down the wrong path to begin with. How many others are like this who have just been riding a hype wave?
Nevertheless, I will say that learning to code is still a useful skill for anyone, even in a future saturated with AI.
Microsoft just laid off thousands and yet are still hiring thousands of H1Bs.
Just one example.
The majority of Adobes jobs on their career page are for foreign workers.
A.I. is misdirection
It shocked me to learn my midsized Southern city (<0.5M MSA) has 1762 H1B jobs, ostensibly because there aren't any qualified Americans; the most common complaint I hear from my peers is that "there aren't enough tech jobs here" when the reality is over 1% of our workforce is underpaid H1Bs, primarily tech field.
I don't think its hateful to want your healthy society's workforce to be made up of citizens, whatever those criteria be determined.
>A.I. is misdirection
It is playing a most useful role for idiots.
66% in silicon valley
Either way, it's AI that's changing the calculus on whether a Georgetown grad who worked her but off to get in, and to graduate with a CS degree, has lots of highly-paid prospects.
My friends in India are all doing swimmingly well with the extremely huge increase in offshoring jobs to India. While in the 2000s it was tech support and in the 2010s it was auxiliary tech workers, today it's actual engineers, UX designers, product guys, even HR and admin functions. Turns out, everything can be sent off to India and LatAm, except for sales and department heads. While previously companies would hire right away from IITs to ship them to the US, today they're content with keeping them in India and letting them work remotely.
The impact of AI in corporate jobs is actually very minimal, but the use of it as a smokescreen for downsizing is uncountable.
Whether it's offshoring or AI-shoring, people's expectations for the long-term effects on the workforce and industry know-how sound interchangeable anyway.
One aspect of college is that it's supposed to teach you how to be a better thinker. One could make the argument that someone who graduated through college should have developed the critical thinking skills required to teach themselves another skill that IS in demand.
What do you tell new grads who can't find a job to do? Learn plumbing, heating/AC, or electrical work? Start a small business?
The whole point of college is to get a desk job and to avoid manual labor.
Plumbers have to literally scoop poop out with their hands once in a while.
Because they can't find a job in the area they studied.
* most colleges
Maybe goodbye to six figure tech jobs at fancy company X, just take the high five figure tech job where you'll have to write a lot of boring C# but at least you're getting paid
I don't doubt that it is much harder to get big salary entry-level work in programming these days. My guess is this is due to a combination of high interest rates / lower investments, flattening business curves, and AI, but the article doesn't try to make a causal case. It just puts forth a bunch of anecdotes sprinkled with a few facts and leaves the reader to infer the causes.
The (3rd party) unemployment numbers do not lie though and the comparison to art and history majors is due to the fact that these degrees (at the bachelor level) usually do not lead to good job prospects, except maybe in roles unrelated to those fields where there is in fact no advantage whatsoever versus another degree.
Every single other engineering field has gone through it. "Regulations are written in blood".
MCAS failures was not a failure of software per se, but a clear system engineering and management failure, and a failure of all engineers involved, including ones that actually are licensed.
If nothing else, MCAS shows the limits of regulation, particularly in the failure mode of regulatory capture (FAA delegated too much power back to Boeing).
