Ask HN: What is the current trajectory of a senior engineer?
Given that LLMs are RL'd to the max with everything that can be given a wrong/right answer (code/maths), what currently is stopping Anthropic, OpenAI et al. from internally building (and maybe buying) datasets of good and bad architecture, and RL'ing that?
Given that LLMs were a fancy autocomplete back in 2024, and now people are talking about building "software factories" with the agents, what in your mind is the stop to this development? Why software architecture would be hard for agents still in a year or two?
Just thinking because I see more and more examples of software developers ignoring this and saying "humans still have a place in guiding the agents and designing the architecture". I think (businesswise) the need for actual software programmer skills is completely out in a few years I think, and product lead etc people will be favored more as they have more people skills.
it was a good run.
71 comments
[ 0.16 ms ] story [ 19.8 ms ] threadWhat will happen is you will get less developers (including less seniors and staff-level) but there will always be expert humans that will check the code before release.
Maybe it was a good run for some here, but you need to look outside of the tech bubble. Do you want an LLM to autonomously fly a commercial plane without any pilots in 2030?
"Seniors" should already know this, but are quickly startled because they are unable to adapt since they based their entire identity and careers around the over-engineering of "clean code", "test driven development" and "architecture" too early on products that make no money, when the LLMs have now cheapened that and taken that role. Even Uncle Bob does not read the code Claude writes in 2026.
Instead, think like an owner.
"Instead, think like an owner." exactly, the traditional software engineer role has ceased to exist and now you must think like an indie entrepreneur and own things. There will be less than 10% of current engineers checking the code in the near future.
> Do you want an LLM to autonomously fly a commercial plane without any pilots in 2030?
No but I also didn't want LLMs to make decisions for the defense department in USA yet here we are.
It will still exist, but far less of them and it includes the reduction of SWEs from junior to staff-level.
Someone will always be held liable if an LLM runs autonomously and out of control which will cost them a lot of money for damages.
> No but I also didn't want LLMs to make decisions for the defense department in USA yet here we are.
There was someone else (another human) that made that decision to use LLMs there, and why certain countries don't trust the US any more, which is my point with both liability and trust in highly critical situations.
AI still isn’t a magic bullet either. I spend some of my days undoing the unholy mess that generated code is.
Exactly. LLMs make it, I'd say, _possible_, to own way more parts of the code than before, and understand even more.
This argument has been made before LLMs and it didn't make sense to me then either. Sure I do if it crashes statistically less often than humans!
But sure, if my job now is going to jail when something malfunctions, that's still better than seeing my family starve.
That is a question I have been asking myself. It was partially answered with Fable 5.1 and Astra. I worked with them on architectural design documents, and they were very good. Some of their suggestions were better than my solutions. They are not at a point where they can fully replace a system architect yet, but they're getting close.
There is still the open question of what to build and how to create a user interface that is understandable and presents complex information in ways understandable to humans. That's something AI cannot do yet.
I suppose eventually it could be a rack robot instead, why not.
I also see a bunch of skills emerging. First if you're doing frontend type dev, there is a lot of value in good aesthetics and UX intuition. You can build hundreds of prototypes and test concepts much better.
On the backend, while LLMs can trace race conditions, the need to reason against spaghetti and figure out what your context window looks like is still very much there. Designing tests that ensure seamless migrations still requires a degree of human oversight.
Then there are frontiers like robotics where even with LLMs coding, there are so many cross functional problems that you dont feel like you are moving faster.
I work in a fairly niche role. I use llms for coding all day but there are times for the type of operations research work I do I simply find it easier to write algorithms in almost correct code.
This statement resonated with me. Thank you.
Human understanding + quality/maintainability is where the money is at. Learn to make those things with agents.
I don't think engineers will truly go extinct, though there may well be less of them. I think the recent discussion around "AI usage is a skill" at https://news.ycombinator.com/item?id=50063427 captures my feeling - there are things to learn, and likely will continue to be, to get the best out of these tools. That means some people will be paid to do it. See to it that you are one of those
Actually, the poetry thing is probably doomed too.
So maybe.. interpretive dance?
There does _seem_ to be some pressure on the top end of SWE salaries. Its possible that this is just wishful thinking, but if true, I suspect that it has more to do with an increase in job applicants rather than a decrease in actual work. AI seems to have driven up the amount of applications received for every vacant position.
If you can direct AI and produce more value per $/time spent then a non software engineer directing AI to develop software then you still can produce a lot of value. This should be the case now for most things that are not a prototype. The job has changed from "writing code" to "shipping software". A technical lead.
software products making ever more intelligent AI software generation accessible will take this job away in many places but there’s also going to be 1000x more software. So there may continue to be opportunities for those that adapt.
Other disciplines have way less publicly available training data than SW.
In other disciplines, the secret sauce of whether a technical decision is good or bad lise within the institutional knowledge and decades of experience of gray beards, not a compiler output that can be integrated in a machine's feedback loop.
The agents have incredible capabilities to develop any bit of code I need and to write good scripts to catch the most basic failure modes, they can even "play the game" by spinning up a VM and using ahk to trigger game state and take screenshots and gifs and use that to understand what it built.
Knowing what to build intuitively to maximise fun, or any other illogical human experience won't be RLable ever. Unless we're at brains in glass jar territory but I don't think we have to worry about that now.
An agent could do this with the SOTA model capability that existed yesterday. But then they’d need to become seriously misaligned.
The problem with AI alignment is that they are only superficially aligned. You can see this difference in people, too. Between the people-pleaser at the office, and the one who says hard truths that really need to be said (but that nobody else is willing to say), which one is really aligned?
And that's literally one single profession.
It's not like engineers didn't know that project management, voice of the customer, and organizing tasks as an intermediary between engineering and management was a thing that could/should/needed to be done. It was just a delegation of tasks before because there was more work than one person could do. Now, all of these roles can be partly offloaded to LLMs, but still having a human in the loop to guide engineering and interface to the business/customer will still be valuable.
Knowing what to build overall, what to build next, what's coming later, and how the product relates to the business goals is still very much a human endeavor.
We're basically turkeys voting for Christmas.
We're cooked.
The economics and the reality are unproven and opaque. Most of testimony we have is heavily invested folk and/or people paying the currently low discounted rates for the technology.
The companies providing the technology are not financially healthy, are heavily geopolitically aligned and a risk for many. If it works or not is moot: it looks like a large economic or political failure on the horizon.
Ergo wait and see before thinking about the future.
The one place where humans will persist are sign-off. Management will want humans to blame. You don’t need to understand necessarily to sign-off but it sure can be helpful.