Done for you
Founding Hires is a search firm for AI research and engineering roles, paid on success. Most of the people we introduce weren't job hunting, and in a lot of cases nobody had approached them yet. We start from our own index of the published research, which tells us who is doing the work and when they come free.
Why teams call us
The researchers worth hiring are usually reachable for a short window: the months around finishing a PhD, wrapping up a project, or watching a lab reorganise around them. Miss it and they are inside a frontier lab on a package you cannot match. We spend our time working out when that window opens.
Every search starts in the literature and on GitHub. We look at what someone has published and shipped, which tells us more in ten minutes than a CV does in a week, and it means we can talk to them about their work rather than about a job spec.
Our sourcing is built around catching people before the recruiting machine does. That is worth real money to you: earlier conversations, less competition on the offer, and candidates who are considering one role seriously instead of five at once.
Short lists here are short. If we cannot find the right people for a brief, we tell you early and stop billing your attention, and sometimes we will tell you the role as written is the reason the search is stalling.
What we hire for
A research scientist and an inference infrastructure engineer share an acronym and not much else. They read different signals, care about different things and need a different pitch, so we run each of these as its own search.
People pushing on pre training, post training, alignment, evaluation or reasoning, with the publication record to back it up.
The bridge profile. They turn a promising idea into training runs that finish, experiments that reproduce and code other people can use.
Engineers who get models in front of users and keep them there: agents, retrieval, tool use, evals, and the cost and latency work nobody demos.
The people who own the GPUs, the serving stack and the throughput numbers, and who keep an inference bill from quietly eating your runway.
Broad builders for AI native startups, comfortable when the spec does not exist yet and happy to own something end to end.
Leaders who can set an AI roadmap, hire the team that delivers it, and stay close enough to the work that good engineers respect them.
You pay us when someone signs. Until then the risk of the search sits with us, which is the arrangement that keeps everyone honest about whether the role is real.
Success based fee, terms agreed before we start
Process
An hour on your stack, your models and what you will compromise on. We calibrate against two or three real profiles before anyone gets contacted.
We map who is doing this work today and who is about to be available, then approach them ourselves, about their work, not with a job description.
A handful of people with our reasoning attached: what they have built, why they fit and where we think the risk is. No CV dumps.
We stay in it through the process and the offer, which is where these hires are usually won or lost, and we tell you when a counter offer is coming.
Contact
Tell us what you are building and who you are missing. If we think we can find that person we will say so within a day, and if we cannot we will say that too.