Datacurve
Research Engineer
San Francisco, CA · Full-time · Posted Oct 30, 2026
About the job
Studies Datacurve’s data and builds infrastructure, experiments, benchmarks, and data strategies to evaluate and improve frontier AI models across specialized domains.
What you'll do
- Design and run experiments end to end, defining data quality standards and evaluation loops for frontier model capabilities.
- Build data synthesis pipelines and feedback mechanisms targeting subtle model failure modes.
- Collaborate with research labs and internal subject-matter experts on data strategies, and deliver research artifacts such as papers, benchmarks, or technical narratives.
What they ask for
- Strong evaluation design, benchmark creation, and error-analysis skills.
- Ability to rigorously investigate what is valuable, informative, and safe for frontier AI models.
- Exceptional communication skills, including synthesizing technical findings into clear narratives and whiteboarding ideas.
- Independent, curious approach to research and problem-solving.
AI focus
Evaluates frontier-model capabilities and failure modes, and develops data strategies, synthesis pipelines, and feedback mechanisms to improve model learning.
Model-written summary · the original posting is authoritative