Stanford AI Expert: These 3 Jobs Will EXPLODE in the Next 5 Years!

Are companies using AIs to replace software engineers?

There's a category of software engineers that I think will be replaced. Hands-on

keyboard, cranking out tickets, it's pretty easy to outsource. [music]

This is Mihail Eric, a Stanford lecturer and head of AI. And today, he's sitting down with us to talk about where software engineering is actually headed.

And while Mihail explains how some engineers will be on the chopping [music] block, at the same time, he introduces us to a new type of engineer that will be more valuable than ever.

What are the top [music] tech jobs that will explode because of AI?

The first one that comes to mind is a There's a huge one that's [music] picked off.

So, today we covered this new type of engineer, plus we covered the skills needed to enter [music] the top 1%.

What would you recommend to getting ahead of 99% of software engineers?

[music] I would spend a lot of time becoming I think it'll be increasingly important.

So, there's a lot of value to unpack in this video, and let's get to it.

Mihail, computer science currently has a unemployment rate of 6.1%, which actually makes it the seventh highest amongst all college degrees. And so, as a person who lectures at Stanford and

interacts with all these computer science [music] students, what's going on? Is the job market cooked?

on? Is the job market cooked?

Because of the surge of coding agents [music] and coding tools and things like this, there's like an upskilling and sort of an up-leveling of the expectations [music] of an engineer from day one. So, where once upon a time you

day one. So, where once upon a time you could enter as maybe an entry-level graduate, and [music] you sort of have this buffer period of maybe 3 to 6 months to maybe a year where you're just kind of, you know, [music] getting your

feet under you. And once upon a time, there was a tremendous expectation to be as productive as maybe the most productive person [music] on the team, you know, the senior engineers. I think

that affordance is no longer as true.

Because of this need to sort of hit the ground running once you actually enter your role, employers are seeing that given two candidates that can come in for the same role, one that is an entry-level and one that is more senior person, we expect that the senior person

will have less of that ramp-up period that we would normally have to account [music] for where we would hire the entry-level person. The employer is

entry-level person. The employer is maybe myopically sort of just prioritizing the thing that is most obvious, which is hey, [music] pick the person who's more senior, they will be able to get more done in a shorter period of time.

The expectation gap from [music] what juniors are to what companies need is the unemployment?

Yeah, I think there's a lot of that at play.

But if I see tools like Auto Code that [music] where I can create literally a a fleet of agents at my disposal that can code, that can plan, that can test, that can do pretty [music] much what a software engineer would do, are companies using AIs to replace software engineers?

There's a category of software engineers that I think will be replaced. If your

only offer to the company [music] is hands-on keyboard, cranking out tickets that are pretty easy to describe in a project management tool like Jira or Linear or whatever a company uses, then that kind of role certainly, it's

[music] pretty easy to outsource, I would say to like a coding agent. So I

will clarify that I think that is something that will no longer be as available as roles. And you know, it turns out that a lot of jobs in the [music] industry do kind of fall into that category. It's typically outsourced

that category. It's typically outsourced labor, it's typically something you'd use like a consulting firm for where they just throw people [music] at the problem. That I think will absolutely go

problem. That I think will absolutely go down.

So diving more deeper into that, if you had to put, what are the top three tech jobs [music] that you think because of AI will disappear?

There is a class of like QA engineers.

Their only real job a lot of the times is to take an existing piece of software and just kind of like test it, click buttons around, and you know, do do that flavor of triaging and and validating on code. A lot of like consulting jobs I

code. A lot of like consulting jobs I think will disappear, like the entry-level consulting jobs I think will also disappear. Usually when an existing

also disappear. Usually when an existing firm will employ like a consulting group, they'll assign them to some task where there's already a project existing, there's already like work that needs to get done. [music] It's kind of like a They just need the manual labor to essentially do that. I think that's something that is increasingly easy to

just outsource to a coding agent.

Entry-level data analysts I think will also kind of disappear where there's just not as much of a need to do rudimentary kind of like data exploration [music] that maybe you'd expect of an entry-level analyst. The

stuff that is quote-unquote not as thoughtful from an engineering standpoint where it is literally you do take a spec, you take a description of of a task, and you just go do that. I

think that entire category of of sort of roles and they may disappear. [music]

Is there even a need for junior engineers going forward if AI is already as good as most of the software [music] engineers?

There is. I think what junior engineers offer uniquely in the engineering landscape, and that I see in in my students [music] is just a tremendous amount of nimbleness in picking up of tools,

picking [music] up of skills, and just being able to tackle problems in a way that at times unopinionated in a good way, right? Because at the end of the

way, right? Because at the end of the day, the world is [music] evolving so so quickly, and where I see seniors even stumble a little bit is just an aversion to change. Whereas a junior engineer,

to change. Whereas a junior engineer, they come in, they have incredibly high slope. If you talk about [music] the

slope. If you talk about [music] the slope versus Y intercept sort of analogy, you know, the Y intercept might not be as high as a senior person, but their ability to like pick up the school tools incredibly quickly is a lot more robust, especially for smaller companies, is something that you [music] want. Cuz small companies, startups,

want. Cuz small companies, startups, they're all about speed, they're all about ability to adapt to problem spaces, to change on a moment's notice.

[music] That's something that I think juniors have, and you know, in their disposal as sort of a meta skill.

Then in that case, why aren't companies hiring [music] more junior software engineers?

Really quick, if you're running a business alone, you're probably spending way more time managing work than actually growing the business. But then,

your whole day turns into doing random admin work that does not move the business forward. That's where Actio

business forward. That's where Actio Work by Alibaba.com comes in. Actio Work

is basically an AI workspace that helps automate the parts of business that usually waste your time. So, as a person in tech, I wanted to see if Actio Work could help me source a new product for my business. I opened the Actio sourcing

my business. I opened the Actio sourcing toolkit, and I asked it to find suppliers for a premium desk [music] mat. Within seconds, it came back with

mat. Within seconds, it came back with supplier cards showing pricing, store ratings, reorder rate, and minimum order quantity. So, I could quickly compare my

quantity. So, I could quickly compare my options instead of spending hours searching manually. After that, I

searching manually. After that, I selected a few of the suppliers and had Actio Work automatically send inquiries and negotiate pricing and MLQs [music] for me. Instead of emailing

for me. Instead of emailing manufacturers back and forth myself, the AI handled the first round of negotiations directly inside the app.

That was honestly my favorite part. Most

AI tools stop after giving you information, but Actio Work actually helps execute the work for you. So, if

you want to easily start up your own business like this, download AxioWork for desktop using the link in my description. Start your 7-day free trial

description. Start your 7-day free trial and unlock the extra bonus credits.

[music] And now, back to the video.

It's a little bit of a short-term optimization. All they're seeing is,

optimization. All they're seeing is, "Oh, no, there's this AI thing that's happening. I'm responsible for reducing

happening. I'm responsible for reducing head count. I have 30 people on my team,

head count. I have 30 people on my team, but I only have this allocation [music] to have engineering resources at my disposal." The only answer is cut

disposal." The only answer is cut people, throw more cloud tokens at it, the solution. Right now, I think just

the solution. Right now, I think just employers are being a little too short-term thinking in terms of how they approach the problem that they're seeing. [music]

seeing. [music] Back in 2021, if I had a computer science degree and I had a pulse, I could probably land a 200K job at Google. And so, obviously, spending

Google. And so, obviously, spending $50,000 on my computer science degree was worth it. Now, in 2026, I can spend even 200,000 on a computer science bachelor's, master's, but still, [music]

when I graduate, I might end up jobless for 6 months. So, in that sense, do you think it's still worth it?

Yeah, I think that again, if your only expectation is to walk in, learn some Python, some JavaScript, some Java, [music] and walk out with a job, I think that is not the right framing of what you're putting in and what you're

expecting to get out. But, if you're going to go and say, "Hey, I can learn how to approach an open-ended project, scope it out, learn how to talk to people, learn how to define a set of like requirements." Emphasizing those

like requirements." Emphasizing those skills, I think college is a unique opportunity, and the degree is a unique opportunity to have a lot of time to kind of dial in on those meta skills that you wouldn't have otherwise.

So, [music] if you had to speak in a broader sense and give maybe like one or two pieces of advice to a software engineer or computer science student, [music] how do you prevent AI from replacing them?

I would spend a lot of time becoming interdisciplinary in a lot of ways. One

of my students, she was someone with a strong UX/UI background. And so, she basically built the system that could translate from a lot of the UI/UX folks, they spend a lot of time sketching. Or

when they're thinking of something, conceptualizing something, is they have a notebook, and they'll draw things out, rough sketches. And she built the

rough sketches. And she built the basically built the system that could take her sketches, you know, meet her where she is, and immediately translate that into literally an app hosted on Vercel, [music] right? A like fully functional web app you can interact with

and the fidelity of the app compared [music] to her sketch was just crazy good. Students that strong in design

good. Students that strong in design that want to combine that with the coding agents. They basically took

coding agents. They basically took [music] two disciplines and they merged them. Folks that are really good in

them. Folks that are really good in hardware systems and they combine that with the coding agent and then they would build something that was specific to the e-discipline, right? So,

combining of multiple things, multiple roles and responsibilities, design, PM, customer success. [music] You know, like

customer success. [music] You know, like all these things could sort of becoming an end-to-end functional responsibility.

I think that's one really big thing that I think is unique and agents don't really do very well. It's all that context that comes from the world.

Okay, so we're talking about becoming a software engineer, but we don't want to just become a regular software engineer.

We want to become a top 1% software engineer. We want to get ahead of 99% of

engineer. We want to get ahead of 99% of software engineers and with the power of AI and agents, [music] we have that.

What would you recommend to getting ahead of 99% of software engineers?

One, they have to be really good at thinking about problems in sort of piecemeal fashion. [music] Uh there's a

piecemeal fashion. [music] Uh there's a second piece which is just knowing how to context switch among different parts of the problem and amongst agents.

[music] And then the third piece is just knowing how to develop of agent-friendly bases. So, just drilling on the first

bases. So, just drilling on the first point of learning piecemeal how to break [music] down a problem. The best

engineers are the ones that know how to take a very complex task and then break it down to the subtasks [music] that need to get organized, the dependencies need to be translated, and then you can from that place kick it off to agents

and have agents [music] do that in in different ways. The second piece and

different ways. The second piece and skill I think is super crucial there is just this concept of knowing how to context switch among different agents.

I've done this work of breaking down the task. Now, first I start by assigning

task. Now, first I start by assigning one chunk of work to one agent. Spend

the time to make a plan for that agent.

The agent [music] is doing his work.

Now, I understand that because of these, you know, that the task decomposition that I've done, I understand that I can now iteratively start to kick off to other agents those other pieces of work that I developed. [music] It's so knowing what's that right point to switch to another agent to kick off that

work. Okay, now you have two agents

work. Okay, now you have two agents doing work. They're kind of doing their

doing work. They're kind of doing their stuff, shipping code in their own way.

Now, what's the point where you can go to a third [music] agent and start like again adding a task into the bucket of things to get done. Knowing how to and go from one to the other and being able to do [music] that seamlessly, now that you have the multiple agents doing work,

is a pretty complex skill because it's it's a bit like psychological skill. You

have to know what's going on with this agent, where [music] this agent's stuck if at all, where do I need to add more clarifications? Oh, now I need to switch

clarifications? Oh, now I need to switch this other agent. So, you know, there there's just that context [music] switching piece that I'm talking about is super crucial.

You're the scrum master of a team of agents.

You're scrum master, exactly. You're

you're the manager of the team of agents, and you have to maintain all that context because they are the most eager interns on the planet. They will

do exactly what you want them to, but they need that supervision. [music] The

third skill I think is super crucial here that I think is not talked about enough is this concept of agent-friendly codebases and agent-friendly development environments. What I usually say to

environments. What I usually say to folks is you want to set up your agents for success. [music] How do you do that?

for success. [music] How do you do that?

Well, you want to ensure that they have a strong test coverage in your codebase.

You want to ensure that they have a strong build system. Basically, the

house in a lot of ways is built before someone goes and enters the house and starts [music] trying to add on top of the house or add a room or or what have you. But, if you do that, that is the

you. But, if you do that, that is the point where you can confidently go off and have a multiple agents [music] go go confident that they're not going to create a sloppy code, go confident that they're going to go and actually [music] confidently build something in the

context of the code that you've developed, the foundation you've set up.

What are the top three tech jobs that will explode in the next 5 years because of AI?

The first one that comes to mind is um a forward deployed engineering. I think

this is this is a huge one that's picked off. The idea of an engineer that

off. The idea of an engineer that essentially goes on site a lot of times with some company or some customer that a company is dealing with and is responsible for translating customer demands or needs [music] into some piece of software using

whatever platform you're selling to the customer. So, it's an interesting

customer. So, it's an interesting [music] job discipline because it is interdisciplinary, right? It combines

interdisciplinary, right? It combines that that sort of PM-ing on-the-ground support work with an understanding of your own codebase.

You're doing the decomposition.

You're doing the decomposition, you're doing the translation, you're the one who's saying, "Hey, the user is really saying this button doesn't work, but what they really mean is there's a whole different feature that they didn't even know they need." They're picking up on maybe the wrong particular detail. So,

the forward deployed engineer, they're on site, they're the voice of the customer in a lot of ways, and then translate that to engineering. The

second and third role, which are related but I think are distinct enough, [music] which is an AI platform engineer and just a core AI feature engineer, right?

So, an AI platform engineer, in my mind, is someone that is essentially building out foundation for those [music] agents to be built, right? So, who's making the decision on how we evaluate, are using [music] Brain Trust, are we using Judgment Labs, are we using like any of

these foundational pieces of software?

What are we using for inference? If

we're actually going to host our own model to run [music] these agents, are we using Fireworks, are we using Modal, or, you know, any of these providers.

So, who's building the platform infrastructure foundations upon which a lot of these agentic capabilities need to [music] be built. So, that's the responsibility of the AI platform engineer.

So, kind of like the AI IT department.

[music] Yeah, in a lot of ways. Now that you have this foundational infrastructure layer that exists across platform in your own company, let's say, how do you then build bespoke [music] features, agentic features on top of that? Now

that you have this foundation built out, I'm building an agent that can do summarization really well, that is grounded in [music] PDFs that are specific to insurance, let's say you're in the insurance vertical. So, there's a whole class of now agent-specific

features that need to [music] be built that an AI feature engineer is the one that is mainly responsible for.

So, as we're closing off, do you have any final pieces of advice for the [music] audience?

If you are in the industry, if you are junior engineer, if you're someone that's entering the workforce, there's an understandable anxiety that a lot of people feel. [music] And what I would

people feel. [music] And what I would highly encourage people to not fall into the trap of is evicting themselves from the entire discipline of [music] software, thinking it's dead, it's cooked, I shouldn't even be a part of this. There is absolutely a role and

this. There is absolutely a role and responsibility that you juniors I think are uniquely [music] poised to fill in this future workforce, agentic development in an AI workforce. You have

to be really nimble, ability to adapt [music] new tools, ability to be very interdisciplinary, and to be able to merge across multiple disciplines in interesting ways. [music] Focus on those

interesting ways. [music] Focus on those unique skill sets before you say that there's no hope here. It's a wrong prediction to say that there's no [music] hope for juniors at all in this new world. It does require a more maybe

new world. It does require a more maybe thoughtful deliberation around where you [music] fit into all of this, but it is one that is definitely still crucial.

We're hiring here in my company, and and I'll gladly you know the right kind of junior folks I'll gladly bring them on.

Cool.

Thank you so much.

Absolutely. Yeah.

Well, that's about all I have in this video. I really hope that you guys

video. I really hope that you guys enjoyed it. And if you did, make sure to

enjoyed it. And if you did, make sure to hit the like button. Subscribe if you haven't already. If you're interested in

haven't already. If you're interested in my absolutely free techniques letter, link for that down below in the description. And if you're interested in

description. And if you're interested in watching the interview that I had with the CEO of Microsoft, Satya Nadella, you might want to watch this video right [music] here.