About the role
ABOUT THE ROLE
Join the founding research team at an AI company to investigate whether expert-authored datasets improve frontier language models and inform future product direction. The role focuses primarily on hands-on experimentation, with collaboration across clients, partners, and the data team.
WHAT YOU'LL DO
- Post-train models on curated datasets, measure performance changes, and diagnose cases where the data does not improve results.
- Design and run evaluations of frontier language models across expert domains such as law, medicine, and finance.
- Select and apply training approaches, including supervised fine-tuning, reinforcement learning, and preference-based methods.
- Turn research findings into dataset specifications and product recommendations.
- Share results through research papers, public benchmarks, and blog posts.
WHAT WE'RE LOOKING FOR
- A PhD in machine learning or NLP, completed or near completion, or an equivalent research record.
- Hands-on experience with reinforcement learning environments for LLMs or RL-based LLM post-training, supported by research, code, or lab or industry work.
- Experience personally post-training and evaluating LLMs end to end, including SFT, RLHF, RLVR, reward modeling, or preference optimization.
- Experience designing or building robust LLM evaluations or benchmarks.
- Early-career research experience, from a PhD nearing completion to a few years post-PhD.
- Ability to communicate technical findings to research, operations, and non-technical audiences, and fluency with AI tools and coding agents.
- Professional English; publications at leading ML or NLP venues or evaluation tooling experience are a plus.
COMPENSATION & BENEFITS
Competitive salary, equity, and relocation support are available. Salary range and visa sponsorship details were not provided.
LOCATION
Full-time, on-site in Barcelona, Catalonia, Spain. Relocation to Barcelona is expected if you are not already based there.