Home / Careers

Small team.
Real work. No busywork.

There are six of us. That means the person who trains a model also watches it fail in production, and nobody's work disappears into a process.

Open roles

No advertised roles right now.

We would rather say that plainly than keep a list of vacancies open to collect applications we have no intention of acting on.

That does not mean don't write

If you work on generative video, audio models, inference performance or GPU infrastructure, we will read what you send. When a role opens, it usually goes to somebody already in that pile — so the useful time to write is before it exists.

Send an open application

The work

What people here actually do.

01

Research that ships

Model work that ends in a render path other people depend on, not a notebook and a result nobody can reproduce.

generative video · audio · inference
02

Infrastructure that absorbs spikes

GPU capacity and cloud rendering on AWS, sized so a catalogue drop is a capacity decision rather than an outage.

AWS · GPU · MLOps
03

Product that hides all of it

A prompt box, a queue, and a finished file. Everything above has to disappear behind one sentence.

full-stack · product
The process

Four steps, no tricks.

  1. 01

    You write

    What you have worked on and what you want to work on. A link beats a CV.

  2. 02

    We talk

    A conversation with someone who does the job, about work you have actually done.

  3. 03

    A real problem

    Something we genuinely have, timeboxed and paid if it takes real work. No puzzles.

  4. 04

    A straight answer

    Yes or no, with the reason. You will not be left waiting to hear nothing.

Think you'd fit?

Tell us what you have built and what you would want to build here.