Work that speaks for itself
Something you built that we can open and look at. A repository, a shipped product, a system still running somewhere. One real thing tells us more than a page of experience.
CareersBuild the Future of
Build the Future of
Intelligent Automation
Join a team of world-class engineers and AI researchers building the next generation of intelligent automation systems. We're solving hard problems at the intersection of AI, infrastructure, and enterprise software.
HiringNo open roles,
No open roles,
but the door is not shut
We are not recruiting at the moment. When that changes it appears here first. If you think you should be working with us anyway, write to us: we keep every application and an engineer reads it. Architecture gets argued out in the open here, and being talked out of your own design is a normal, good outcome.
Steady when the deadline is real
Not fast. Steady. The difference shows up in the week something breaks the day before a launch, and it is the single hardest thing to interview for.
The same standard in month six
Anyone can be careful for a fortnight. We are looking for the person whose tenth pull request is written as carefully as their first.
Owns the outcome, not the ticket
Sees the thing through to working in production, says so early when it will not land, and does not need to be asked twice.
If that reads like you, send your CV to [email protected] with two things: which of the four above you are strongest on and what you did that proves it, and what you would want to work on here. A paragraph is plenty. We answer either way.
How It WorksIf We Think
If We Think
There Is a Fit
Nothing here is a queue you enter by emailing. If what you send lands, this is what follows: conversations with the people you would work alongside, not a panel of strangers or an algorithm quiz.
You Email Us
Your CV or GitHub, and a short note about something you built that you are still proud of. Cover letters written by a language model are obvious and do not help you.
Technical Conversation
Ninety minutes with the engineers you would work alongside, going deep on something you already know well rather than testing whether you memorised an algorithm.