The problem we're solving

The first era of AI was built on information that already existed: text, images, code, and the rest of the internet's digital record. The next generation of intelligent systems will need something different. Intelligence depends on context, lived experience, human judgment, and an understanding of how the physical world actually works. Most of that knowledge isn't sitting online waiting to be downloaded. It has to be created.

Today, the systems that produce this kind of data are fragmented, expensive, and hard to scale. Human experience is continuous, diverse, and always changing, and the infrastructure that captures it should work the same way.

That's why we're building Kaius: the human data infrastructure for AI model training. We've built an always-on, global network where people contribute real-world experiences, perspectives, and knowledge to the development of intelligent systems. Our platform lets AI teams collect exactly the data they need, by language, location, and person, across voice, image, video, and text. Participation is engaging, gamified and rewarding, and the network adapts as the needs of technology evolve. Just as important, it's something people actually love doing.

We believe that if you're going to build a product, you should build one hell of a product. That's why our small team of repeat founders and builders are backed both by venture, and by some of the best product and technology minds, including the Cofounder/CTOs of Slack and Duolingo, Reddit's former CEO, and senior leaders at top labs.

What you'll do

• Own data collection program execution, quality, margins, and revenue delivery.

• Lead the solutions side of every new customer conversation

• Write the scope before we sign: who we collect from, where, how much, the rubric, the quality bar, the timeline and the price

• Work with engineering and data operations so what we promise is what we ship

• Bring patterns back to product. If three customers ask for the same thing, that's a roadmap item

• Build what makes it repeatable: scoping templates, quality frameworks, pricing for custom work

• Build out a rockstar solutions team underneath you

Your first 90 days

You'll own scoping on every new deal, and customers will start asking for you by name. You'll guide the current collection team on current efforts and support both existing customers and new one.

What we're looking for

• You've been on the provider side of AI data. You've built training data programs for model builders, not just bought them. This is the one we won't flex on

• You speak pre-training and post-training. You can read a paper and say what data would move the needle

• You've written scopes and proposals that turned into real, delivered work

• You're a builder. A pricing model, a quick tool, a rubric. You don't need a team to get started

• You write simply. We're remote-friendly, and concise writing is how we think clearly

Nice to have

• Your specialties are in voice and audio data, multilingual collection, robotics or egocentric video

Probably not for you if

You need a final-stage product and a massive delivery team behind you before you can move. Or your best work ends at a great demo or a great deck. We need someone who's shipped the thing before, and wants to do it again, better.

Why now

Hard, valuable, fun. The problem has no playbook, the market is wide open, and the team already likes each other. You'll be the first person in this seat, and you'll help shape what it becomes. This is what you'll be known for.

How we hire

Selling to frontier labs and F500 teams means being in the room, so this seat sits in San Francisco or New York. The rest of the team is remote-friendly. No AI screens, coding challenges, etc. A couple of hours working with us instead.