Why This CEO Stopped Caring About PyTorch on Resumes

“Do I care if someone knows PyTorch or knows how to program CUDA kernels? I don’t care. AI will be better at writing all that code than you in a year anyways.”

That’s Wiley Jones, CEO and co-founder of Doss, the AI-native operations platform that just raised $55 million for mid-market physical operations companies. He’s hiring engineers right now. And the things he’s looking for have almost nothing to do with what most job descriptions list.

The shift that already happened

Jones sees a parallel between what’s happening now and what happened during the SaaS era. In the SaaS years, software engineers didn’t have to think deeply about the technology they were writing. The infrastructure was abundant, the applications were thin, and the product mattered more than the engineering underneath.

“AI engineers today don’t have to think about the AI,” Jones says. “They think about the systems engineering and they’re not thinking about how to build a really good PCA and regression classifier. They’re instead thinking, how do I collect the context and build a system that can go out and take action in the world?”

The label on the resume still says “machine learning engineer” or “AI engineer.” But the actual work has moved. The person Doss is hiring right now worked on self-driving cars and fine-tuning — traditional ML infrastructure. What makes him valuable isn’t those specific skills. It’s that “he’s a systems thinker.”

What the interview actually tests

Doss’s interview process sounds simple on the surface. Jones describes it as “a lot of reasoning” — give the candidate a problem, work through it together, observe how they think.

But the specific things he’s evaluating are precise: “Can they hold a complex idea in their head? Can they break it down into pieces? Can they reason about each piece and then sequence them? Can they deal with hard invalidations where you’re thinking about the invariance of something and then how they conflict?”

Notice what’s missing: no whiteboard coding, no framework-specific questions, no “implement a linked list.” The evaluation is about cognitive architecture — how someone structures a problem, not what language they write the solution in.

“I don’t actually care about the code,” Jones says. “I care about how you think.”

The bomb diffusion test

Jones uses a thought experiment that captures what he means by “systems thinking” applied to product problems. Imagine an AI agent that knows the correct answer to a question but doesn’t have authorization to share it. The information is protected. But sharing it would prevent harm.

“Do you tell them the answer?” Jones asks. “It’s like, well, they can’t actually know this because they’re not allowed to see it. But I do know the answer. So I can either tell them, you’re not allowed to know the answer, or do you tell them, I’m not even sure what you’re talking about?”

This isn’t a machine learning problem. It’s a product problem about progressive disclosure, access control, and ethics. The kind of engineer Jones hires isn’t the one who can optimize the model — it’s the one who can think through the system-level implications of what the model knows.

“The way that machine learning engineers think about some of these things in terms of progressive disclosure and training — these are concepts that these systems thinkers are really good at applying themselves into using the substrate and framework of their academic background.”

Fewer people, better paid

The hiring philosophy extends to team composition. When asked whether AI changes how he thinks about team size, Jones gives a one-line answer: “You want to hire better people, pay them more money, and have less of them.”

This isn’t just about efficiency. It reflects a belief that the leverage per person has increased so dramatically that the constraint is no longer headcount — it’s the quality of judgment each person brings. One engineer with strong systems thinking and genuine creativity generates more value than three who can write clean code but can’t reason through the product implications.

Jones’s own behavior backs this up. He’s the second or third highest Claude Code user in his company. He brain-dumps into ChatGPT on his commute, processes it through Pro mode, then hands it to Claude Code for execution planning. The tools amplify judgment. They don’t replace it.

FAQ

What skills should AI engineers have in 2026?

Creativity, agency, and judgment matter more than framework knowledge. Wiley Jones says AI engineers today don’t need to know machine learning — they need systems thinking: how to collect context, build systems that take action, and reason through product implications. The ability to hold a complex idea, decompose it, and sequence the parts is the core skill.

How does Doss interview engineering candidates?

Through reasoning exercises, not coding challenges. Candidates work through problems with the interviewer, evaluated on whether they can hold complex ideas, break them into pieces, sequence the parts, and handle hard invalidations. Jones says “I don’t actually care about the code. I care about how you think.” No framework-specific questions.

What is Doss and what kind of engineers does it hire?

Doss builds an AI-native operations platform for mid-market physical operations companies, serving brands like Verve Coffee and Eight Sleep. It hires systems thinkers — engineers who can reason about product-level problems like access control, progressive disclosure, and multi-agent coordination, regardless of whether their background is in ML, self-driving cars, or traditional software.

Should engineers learn PyTorch or CUDA in 2026?

Wiley Jones explicitly says he doesn’t care if candidates know PyTorch or CUDA kernels — “AI will be better at writing all that code than you in a year anyways.” Framework knowledge depreciates fast. Systems thinking, structural data modeling, and the ability to reason about product implications are more durable and harder for AI to replace.

How is the AI engineer role different from machine learning engineer?

ML engineers focused on building classifiers, training models, and optimizing inference. AI engineers in 2026 focus on systems engineering — collecting context, building agent systems that take action, and solving product problems like situational access control. The academic background is similar, but the work has shifted from model optimization to system design.

What does “hire better people, pay them more, have fewer of them” mean in practice?

With AI amplifying individual leverage, one engineer with strong judgment creates more value than several who can only execute. Jones’s team uses AI tools extensively — he’s the second or third highest Claude Code user in the company. The tools multiply the output of good judgment but can’t replace it, making the quality of each hire more important than headcount.

How do you test for systems thinking in interviews?

Present problems that require reasoning across product, ethics, and engineering simultaneously. Jones uses thought experiments like situational access control — should an AI reveal protected information to prevent harm? The answer isn’t about technical implementation. It tests whether the candidate can think about progressive disclosure, competing constraints, and system-level trade-offs.

Is academic background still relevant for AI engineering roles?

The background matters less than the thinking patterns it produced. Jones’s current hire worked on self-driving cars and fine-tuning — traditional ML infrastructure. What makes him valuable is systems thinking, not those specific skills. “A lot of the same primitives and concepts, but applied through a new lens.” The academic framework is a substrate, not the skill itself.

What is the bomb diffusion thought experiment for AI access control?

An AI agent knows the correct answer but lacks authorization to share it. Does it reveal protected information to prevent harm, or pretend it doesn’t know? This tests whether engineers can think about product-level access control, not just model accuracy. Jones uses it to evaluate whether candidates think in systems rather than algorithms.