Founding AI Researcher

About GenosenWe are an AI research lab in Bucharest.Our work is on interpretation: how a defined audience understands a piece of communication, and how to predict that before it reaches them. Our first model, P/26, applies this to marketing artefacts. Later models will cover other domains where the same problem appears.Where we are today: a closed-beta program with testing users, a working system, a pre-seed round in progress at a 4M EUR pre-money valuation, and a validation study planned for later this year that will measure our accuracy against real outcomes and publish the result whichever way it goes.The RoleYou would be one of the first people on the research side of the lab, working directly with the CTO and the CEO.The work is model conception, calibration, and evaluation. Designing the validation methodology, testing predictions against real audience behaviour, and finding the places where the model is wrong before a customer does. This is a research problem with a product attached to it, and you would own a meaningful part of it.We will go into specifics at interview stage.What we are Looking ForAt least two years working on AI itself. Modelling, evaluation, experimental design. Software engineering experience is useful, though it is the AI work we need.A degree in AI, machine learning, computational cognitive science, or a closely related field.Interest in how people perceive things. Psychometrics, cognitive psychology and behavioural science all touch this work.Willingness to be measured. We are building something that makes falsifiable claims, and we intend to test those claims publicly.Compensation, Stated PlainlyWe cannot pay a salary yet. What we can offer is equity in a company currently valued at 4M EUR pre-money, with a vesting schedule agreed before you start. We expect to move this role to a paid position in about twelve months, once the pre-seed round has closed and revenue supports it.That is a real trade and we are not going to pretend otherwise. If the timing does not work for you, that is a reasonable conclusion. Tell us anyway and we will come back to you when we are paying.Who this is forPeople who want to own a research question rather than close tickets against someone else's. People who want their name on the validation paper. People who would rather build the second model in the family than maintain the first one.If that sounds like you, send a CV and a short note about a problem you got wrong and what you changed afterwards.