This Research Product Manager is a research-first product role focused on the model itself, not a traditional application backlog. You will work with researchers at frontier AI labs to identify capability gaps, design evaluations and training-data programs, run experiments, and manage delivery from research scoping through weekly data releases.
What you'll do
• Co-design model capabilities by identifying gaps and running headroom analyses.
• Run experiments and benchmarks using simulation infrastructure and evaluation tooling.
• Translate ambiguous partner requests into research questions, project plans, and aggressive milestones.
• Design data-curation programs using domain experts and synthetic-data generation.
• Conduct literature reviews and implement state-of-the-art research in practical evaluation frameworks.
• Own one-to-three-month projects from scoping through weekly delivery, feedback, and iteration.
• Bridge research and operations across researchers, customers, and external contributors.
Requirements
• 0 to 5 years of experience in AI/ML research, technical product management, research engineering, or another deeply technical role.
• Fluency in LLM evaluations, benchmarks, coding agents, and modern model-development workflows.
• Strong Python skills and comfort with data, experiments, and technical tooling.
• Demonstrated ability to ship technical projects end to end.
• Research-minded judgment, curiosity, and comfort with open-ended questions.
• Excellent communication and project leadership across technical and non-technical collaborators.
• High ownership, urgency, and comfort working in an early-stage environment.
• Ability to work five days per week on-site in San Francisco.
Benefits
• $150,000 to $240,000 base salary.
• Sizable performance bonus tied to project and company milestones.
• Meaningful early-stage equity.
• Health, dental, and vision coverage.
• Flexible paid time off.
• Direct access to frontier AI researchers and substantial ownership over model-development programs.
The company can consider OPT and existing H-1B transfers.
