AI Engineer

Most systems understand the internet after it happens.

We’re building something that understands it before.

Vote is building the attention signal layer of the internet. A system designed to detect early momentum across content, narratives, creators, and markets before consensus forms.

This sits at the intersection of:

AI

consumer social

prediction markets

on-chain data

gaming mechanics

and cultural trend formation

We’re not building a model.

We’re building a system that captures early conviction and turns it into intelligence.

If you’ve ever thought:

“why do we only measure things after they blow up?”

you’ll understand what we’re doing.

We believe the most valuable signal on the internet is early belief under uncertainty.

Not opinions.

Not engagement.

Not predictions with fixed outcomes.

But the moment someone sees something and knows.

We capture that.

Then we combine it with:

• multimodal content signals (text, audio, visual, context)

• temporal attention dynamics

• user reputation and calibration

• network propagation patterns

• on-chain coordination and incentives

and turn it into a continuously updating attention graph.

This is closer to an intelligence aggregation system than a traditional ML pipeline.  It behaves more like a market than a model. Closer to a sensing network than a product. Over time, this becomes infrastructure.

There is no clean dataset. There is no ground truth. There is no “correct answer.”

We are modeling:

emergence

attention formation

cultural phase transitions

human intuition under uncertainty

Which means:

• timing matters more than accuracy

• weak signals matter more than strong ones

• aggregation beats prediction

• reputation becomes a feature

• uncertainty is part of the system, not a bug

If you want clean supervised learning, this isn’t it. If you want to build something new, it probably is.

This is a founding AI engineer role.

You will help design the core system, not just implement it.

You’ll work on:

• multimodal representation pipelines (text, audio, visual embeddings)

• ranking and scoring systems for early signal detection

• temporal modeling of attention and velocity shifts

• reputation-weighted systems and calibration layers

• ensemble-style architectures across human + machine signals

• noisy, incomplete, real-world data systems

There’s also a strong overlap with:

on-chain data systems

token-incentivized networks

market design and information aggregation

You’ll be working directly with the founding team and shaping the core architecture from day one.

Parts of this look like:

a recommendation system

a prediction market without outcomes

a game

a data network

a social product

a financial primitive

All at once.

We are pulling from ideas in:

collective intelligence

multimodal ML

ensemble systems

information markets

behavioral data systems

and compressing them into one system.

You’re probably someone who:

• thinks most ML systems are too constrained or obvious

• enjoys working with messy, ambiguous problems

• cares about systems, not just models

• has built things from scratch before

• understands tradeoffs between theory and reality

You might have experience in:

recommendation systems

ranking / retrieval

multimodal ML

time-series systems

market-based systems

distributed systems

Or you might not. But you think like someone who could.

How you think

What you’ve built

How you approach undefined problems

We do not care about:

perfect resumes

checklist experience

over-optimized academic paths

Python

modern ML frameworks

vector / embedding systems

real-time data pipelines

some on-chain components over time

This will evolve quickly.

Direct ownership of core systems

A real founding role

Exposure to both consumer scale and data infrastructure

The chance to work on something that doesn’t already exist

If we get this right, this becomes a new primitive.

Start: June (flexible)

Location: Remote + SF/Toronto

• Some travel (US, Europe, Asia)