Research Engineer (YC Start-Up) - Up to £150k + Equity

Few&Far are working with a small, fast-moving YC-backed AI startup that's building frontier systems modelling complex, large-scale real-world behaviour - genuinely hard from both a research and engineering standpoint, not an application layer bolted onto existing models. They're already working with some of the biggest organisations in the world and have already passed the 1M ARR milestone.The roleThey're looking for a Research Engineer to own the path from research to production. Their research team is inventing new methods; you'd take them from prototype to something running reliably at scale - post-training and distilling the models that power them, building the data pipelines that feed them, owning inference at scale, and simplifying aggressively so only what actually delivers performance makes it to production.You'd work directly with the CEO and Head of Research, and partner closely with research, product, and forward-deployed engineering.What they're looking for2+ years shipping ML in a startup environment, ideally at an early-stage company where you owned things end-to-end.Real hands-on post-training experience with LLMs - RL, fine-tuning, distillation, dataset and reward designStrong, clean Python that other people can maintainComfortable reasoning about statistics, probability, and high-dimensional spacesFast, with a high bar - goes deep to understand both problem and solutionGenuine curiosity about the underlying domain, not just the MLOwnership mentality - sees what needs to happen and makes it happenNice to haveExperience with distributed training, inference optimisation, or scaling ML systemsBackground in an empirical field like behavioural science, computational social science, psychometrics, or statisticsPublications or open-source work in simulation, evaluation, calibration, or behaviour modellingIf you hit around 60% of this criteria, apply anyway - we'd rather see strong candidates who don't tick every box than miss out on good people over a checklist.