AI Engineer (AIE) vs. Machine Learning Engineer (MLE) : Why Hacker News Purists Are Wrong

What exactly does 'AI Engineering' mean and how is an AI Engineer different from a Machine Learning Engineer?

Everyone’s talking about AI, but not everyone agrees on what “AI Engineering” actually means — especially when compared to the more established role like Machine Learning Engineer.

Machine Learning Engineers are typically the ones that build the AI, and the job title is a standard one at leading tech companies like Meta, Uber, and Netflix. So at first, it might seem odd that ‘AI Engineering’ largely refers to something else other than what those engineers building AI are doing.

In this article, I’ll explain how I think the terms are evolving, and why I believe “AI Engineering” is going to become the dominant label for a growing category of engineering work. In essence, AI Engineering is becoming a shorthand for Applied AI Engineering, building applications, systems, and tooling on top of pretrained large foundational models — especially LLMs and image generators like Stable Diffusion.

AI - or Artificial Intelligence - has been around since the 1950s so it might seem strange that AI Engineering is such a recent term. But it is, and I’ll explain why below. And like all new terminology, there’s no clear consensus on what exactly it means.

The definition I’ll use in this article reflects a growing industry trend — and one I believe is likely to become the standard, for reasons I’ll explore, even if it contradicts how some purists feel. There's also uncertainty around whether the formal job title “AI Engineer” will catch on, or whether it will morph into something like “Software Engineer — Applied AI”.

Since I’ve named this newsletter AI Engineering Report, it’s only fair that I explain what the term means to me. I’ll also compare how AI Engineering compares to Machine Learning Engineering work.

AI Engineering vs. ML Engineering

The definition of "AI Engineering" varies depending on who you ask.

Some purists argue that only those building the AI models deserve the “AI Engineer” label — and that those merely using the models are just software engineers applying machine learning.

I think the purists are going to lose that fight.

Yes, their argument makes sense in a historical context. But “AI” has become a tidal-wave buzzword, and in practice, it now mostly refers to using models like GPT, Claude, and Stable Diffusion — not building them from scratch. Far more people will work with these models than will train them.

And importantly, the people building the models already have accurate titles: AI Researcher, Research Engineer, ML Engineer. There's room for a new label for those who specialize in building software with these models at the core.

A Brief History of AI Nomenclature

AI has always had many flavors — from brute-force search (e.g. chess), to symbolic logic (rule-based systems), to today’s statistical and neural methods.

Ultimately, machine learning — especially statistical and neural approaches — became the dominant paradigm for real-world AI products.

From Google Search and Translate, to the Facebook feed, to Netflix recommendations and Google Photos — machine learning powers nearly every major "AI product" of the last 15 years.

The people building those systems were called Machine Learning Engineers, because that’s what they were doing day-to-day: training models, optimizing pipelines, working closely with data and researchers.

Meanwhile, most academic researchers simply labeled their work AI research. Even if it included engineering, it wasn't usually framed that way.

Because the industry embraced "machine learning" as its operational term — and the academic community was smaller and less media-visible — the term AI remained loosely defined… until the rise of LLMs and generative models in the 2020s.

Today, AI in popular discourse largely refers to generative foundation models — especially LLMs. So it’s natural that AI Engineering now refers to engineering work focused on using those models.

ML Engineer vs. AI Engineer: What They Work On

Let’s be honest: job titles are messy, and responsibilities often overlap.

Roles like software engineer, data scientist, infra engineer, and ML engineer frequently blur. But we can still draw a rough outline of where things tend to cluster.

Machine Learning Engineers:

These are the people who build foundational models and bring them to production scale. Some specialize in algorithm design and research; others focus on training pipelines, distributed compute, and massive-scale data engineering.

Common areas of work:

• Ranking systems (e.g. search engines, ad platforms)

• Recommendation systems (e.g. Netflix, YouTube)

• Classification (e.g. spam detection, fraud)

• Forecasting and time series prediction

• Computer vision and traditional NLP

Applied AI Engineers (AI Engineers):

These are software engineers who build apps, tools, workflows, and systems on top of pretrained models — especially generative ones like GPT-4 or Stable Diffusion.

Common areas of work:

• Prompt engineering, prompt routing, and finetuning

• Retrieval-augmented generation (RAG) and vector search

• LLM pricing and latency optimization

• Model benchmarking and evals

• Agent architectures

• AI-Assisted coding and AI-Driven coding (vibe coding)

• Apps with LLM technology as a central feature

---

No Neat Boxes

In practice, there's no clean boundary between AI engineering and other disciplines.

A senior FAANG engineer recently told me: “At our company, data scientists run prompt evals, and software engineers handle prompt routing optimization.”

This reflects the reality that AI Engineering often spans teams and disciplines. And while I can’t say with certainty where industry consensus will land, I’m absolutely confident that this type of work is exploding in complexity, scope, and importance.

That means we need better language and clearer mental models — and this newsletter exists to help shape that conversation.

Further Reading and Notes

This article was heavily inspired by Latent Space’s “The Rise of the AI Engineer” article:

The Rise of the AI Engineer

Their research on job postings was especially insightful, where they pointed out that AI engineering job postings are on pace to overtake ML Engineering job postings.

It was also inspired by Vercel’s post, Becoming an AI Engineering Company.

I’m still deep playing with MCP, and plan to return to the topic next edition. In the meantime, check out the Substack ‘MCP in Context’ if you want a newsletter even more focused on MCP. In a recent post, the author explained what entities will win and lose if MCP adoption continues to grow:

MCP's Winners & Losers

I’d love to hear any thoughts on this topic - reply by email or comment directly on Substack.