Ship agents that people actually depend on.
We are small, senior, and deliberately unglamorous about the work. If you want to build AI systems that survive contact with real users — and be accountable for the number at the end — this is the right room.
What the job is actually like.
Written plainly, because the mismatch usually shows up in month two otherwise.
You scope what you build
The engineer on the discovery call is the engineer who ships it. No throwing a spec over a wall, and no inheriting one either.
You own a number
Every system has an eval pass rate and someone accountable for it. That is uncomfortable at first and the reason the work is good.
You will say no to clients
We talk people out of builds that will not work. You are expected to make that call and to make it early.
Deep work is protected
Two meeting days a week. The rest is heads-down. We measure output, not presence in a calendar.
Where we need people.
Apply by email with something you have built. We read every message and reply within a week.
Senior Agent Engineer
Apply →Eval & Reliability Engineer
Apply →Full-Stack Product Engineer
Apply →AI Solutions Architect
Apply →Nothing that fits? Send us what you have built anyway — contact@flashcycle.ai. We have created roles for the right person before.
What we offer.
Four steps, about two weeks.
No take-home longer than three hours, and we pay for it if it goes beyond that.
Intro call
Thirty minutes with an engineer, not a recruiter. You ask as much as we do.
Technical conversation
We walk through something you have built and something we have. No whiteboard puzzles.
Practical exercise
A scoped, realistic agent problem. Three hours maximum, reviewed by the team you would join.
Offer
Decision within a week of the exercise, with feedback either way.
Send us something you built.
A repo, a system, a write-up of something that broke and what you did about it. That tells us more than a CV.