Xterra AI
Founding Research Scientist
San Francisco · On-site · Full-time · Posted Jul 20, 2026
- Experience
- Hiring across early-career through senior/staff levels; no specific years stated.
- Education
- For PhD students, the company encourages completing the PhD; mid-program students are preferably engaged through advisor-supported paid research residencies or collaborations, with full-time roles after graduation.
About the job
Develops and trains AI systems to reason about scientific problems, taking research ideas through experimentation and into production.
What you'll do
- Design and train reinforcement-learning systems for scientific hypothesis generation and evaluation, including reward models and verifiers.
- Research scalable oversight, alignment, and domain-specific evaluation methods for complex scientific reasoning.
- Build training infrastructure and run large-scale experiments across the research-to-production lifecycle.
What they ask for
- Strong machine-learning fundamentals and hands-on experience training large models; LLM experience is preferred but not required.
- Demonstrated reinforcement-learning experience, from language models or another domain.
- Ability to work across research and engineering, including writing papers and debugging distributed training jobs.
- Familiarity with at least some of reward modeling, RLHF/RLAIF pipelines, search and planning methods, or AI alignment techniques.
Nice to have
- Publication record.
- For senior levels: independent research leadership, mentoring, technical-strategy influence, and deep expertise in a core technical area.
Tools
RLHF, RLAIF
AI focus
Research and engineering for AI systems that reason about scientific problems, including reinforcement learning, process reward models and verifiers, scalable oversight, alignment, and evaluation.
Model-written summary · the original posting is authoritative