Inside Athyna Intelligence with Bill Kerr

Welcome to Revenge of the Nerds. We’re skipping the hype and going straight to the builders. In this edition, we talked about:

• The next phase of AI progress won’t be driven by bigger models alone, but by post-training: expert-driven evaluation, reasoning-heavy data, and alignment work that turn systems into reliable, real-world tools.

• The AI industry is hitting a structural talent bottleneck. The U.S. PhD pipeline can’t keep up with growing demand for advanced researchers.

• Athyna Intelligence, a new initiative from Athyna, brings PhD-level expertise from LATAM into AI post-training—built on years of experience hiring, vetting, and supporting top talent across the region.

Let’s dive in. No floaties needed.

Revenge of the Nerds

Bill Kerr, Founder and CEO at Athyna

Bill Kerr is the founder and CEO of Athyna, a global talent platform helping companies hire exceptional talent across the globe. Alongside building Athyna, he’s an active mentor and investor at Startmate, a limited partner at Blackbird, and an early backer of companies including Humane, Future Super, and Atomic8. Through his writing at Open Source CEO, Kerr also shares deep dives on leadership, startups, and the evolving nature of work.

Kerr’s path to building Athyna was anything but conventional. Before founding the company, he spent a decade outside the startup and corporate world—working as a tradesman, investing in real estate, and later traveling extensively across more than 40 countries. That non-linear journey, combined with multiple failed and formative ventures, shaped his belief that talent is global, opportunity is unevenly distributed, and infrastructure—not geography—determines who gets to build the future.

What sparked the idea for Athyna Intelligence? What problem did you see that wasn’t being solved?

If you zoom out, the AI bottleneck is shifting. Compute is still fundamental, it’s one of the pillars, of course, but once research and compute are in place, the real constraint becomes the expert data that shapes post-training.

This is the phase where models learn to reason, follow instructions, avoid failure modes, and become reliably useful, and it is powered by one thing: expert hours, not generic labeling.

The problem is that most AI training infrastructure was built on the assumption that expertise lives in a few expensive cities, forcing companies to fight over the same limited talent pool. It works, but it doesn’t scale.

The reality is that expertise is global. Latin America produces researchers with exactly the domain depth frontier models need. Athyna Intelligence exists to unlock that expertise and build the infrastructure that finally delivers it at scale.

Why do you believe Latin America is uniquely positioned to support the next wave of AI innovation?

Because the future of AI training will belong to the people who can scale expert-hours as fast as they scale compute, and Latin America is one of the only regions in the world with that kind of capacity.

The region graduates tens of thousands of PhDs and Masters students every year across mathematics, physics, computer science, biology, law; the exact disciplines where post-training gets its strength.

These aren’t generic annotators. They’re domain experts who can write rigorous reasoning chains, evaluate frontier model outputs, and catch the edge cases that actually move the needle.

And unlike other talent markets constrained by cost or geography, LATAM has three defining advantages: Time-zone alignment with the U.S., deep academic infrastructure connected to global research institutions, and cost structures that expand what's possible, not limit it.

In other words, it’s a region capable of supplying the expert hours the industry desperately needs, and doing it with scientific rigor.

In simple terms, how does Athyna Intelligence actually work?

Athyna Intelligence starts with a very pragmatic focus: people. AI post-training only works when the right experts are involved, and teams need to access that expertise quickly. Our first step is connecting companies with vetted PhD and Masters-level researchers from Latin America, available on demand and ready to contribute with fast turnaround, bringing the academic depth and analytical rigor this phase of AI development requires.

From there, we will build the operational layer around those experts. That means designing the workflows and execution processes that translate expert judgment into scalable training output, from RLHF and reasoning-heavy annotation to evaluation protocols and quality assurance loops.

We will develop this work in close collaboration with design partners, using real production needs to shape how tasks are structured, reviewed, and iterated. The focus is on building practical, scalable workflows that deliver speed, accuracy, and real-world impact.

Ultimately, Athyna Intelligence is about turning global expertise into reliable, repeatable AI post-training outcomes. We start with people, build the workflows around real production needs, and let the technology follow.

What kinds of AI or data challenges are you seeing in the market that show there’s a real need for this initiative?

When you look at what’s happening across the AI landscape, a pretty clear picture starts to emerge. AI-related roles are among the fastest-growing in the U.S., with AI Engineer, AI Consultant, and AI/ML Researcher now ranking in the top 5 fastest-growing positions nationwide—roles that barely existed five years ago. This tells you something important: companies aren’t just experimenting with AI—they’re actively searching for deeper expertise to push these systems further.

We’re seeing models hit the same limits again and again, especially when it comes to reasoning. Math, logic, physics, causality… these are the places where generic labeling simply can’t push models any further, and you start to feel the absence of real domain expertise.

At the same time, safety and alignment teams are stretched thin. Red-teaming a frontier model isn’t a ‘poke it and see what happens’ exercise. It requires people who actually understand the underlying concepts, who can stress-test a system with intent rather than guesswork. And very few teams have enough of that kind of expertise in-house.

On top of that, evaluation pipelines are becoming increasingly fragile. Everyone knows evals matter, but not many organizations have the talent or the operational infrastructure to run them with the rigor these models demand. The gap between what companies want to validate and what they’re actually able to validate keeps widening.

So the pattern is unmistakable: the industry needs better data, the kind of data that actually improves models. And that only comes from people with real expertise. That’s the shift Athyna Intelligence is built to support.

When you look at the broader market and the main players in this space, what observations shaped how you and your team are building Athyna Intelligence?

We noticed three truths shaping the future of AI training.

First: AI doesn’t get smarter from pre-training alone. It gets smarter from the quality and quantity of expert-hours applied after the model is built.

Second: expertise is the real bottleneck. Most platforms were built on the assumption that expertise lives in the U.S. and Europe. But that assumption is already breaking, the supply just isn’t there.

Third: the industry needs a new kind of infrastructure. Something that can deliver domain expertise at speed, at scale, and at a cost that lets companies reinvest in what matters: better models, deeper research, longer context windows.

Latin America gives us the talent layer, deep academic expertise across disciplines. Athyna gives us the infrastructure: a decade of experience in sourcing, vetting, quality control, and global talent delivery.

Put those together, and you get something the market hasn’t had until now: expertise at global scale, with the economic headroom to deploy it widely. That’s the blueprint for Athyna Intelligence.

How do you see Athyna Intelligence creating opportunities for high-level talent in Latin America that didn’t exist before?

This is the part I’m personally excited about. For decades, highly trained researchers in Latin America had only two paths: leave the region, or stay and work far below their intellectual potential.

Athyna Intelligence creates a third path, one that didn’t exist before. It lets researchers work on frontier AI problems from where they live, earn globally competitive compensation, and apply their expertise to technologies that will define the next decade.

It’s unlocking global participation in one of the most important scientific shifts of our time. And from everything we’re seeing, this is just the beginning.

Additional Reads:

• Bill Kerr’s interview on Open Source CEO Newsletter.

• How Athyna Does Remote — Open Source CEO Newsletter

• If You Ain’t AI-First, You’re Last — Open Source CEO Newsletter

Meme Of The Day

Saturday Poll

Rate This Edition