TL;DR
AI Product Manager, Insights: Owning model evaluation analysis and insight generation for AI systems, transforming raw result data into a roadmap for model improvement with an accent on deep analysis of model failures, semantic patterns, and systemic hallucinations. Focus on structuring findings into a logical hierarchy to deliver critical, data-driven narratives to senior leaders.
Location: Must be based in San Francisco, New York, or Seattle, USA
Salary: $206,800–$258,500 USD
Company
hirify.global develops reliable AI systems for the world's most important decisions, providing high-quality data and full-stack technologies to power leading models.
What you will do
- Own the creation of model evaluation from initial hypothesis, data scraping to final publication.
- Deeply analyze why models fail, identifying semantic patterns, edge cases, and systemic hallucinations in raw model outputs.
- Review raw data sets, meeting transcripts, and research notes to identify key findings.
- Act as the bridge between data and narrative by structuring findings into a logical hierarchy.
Requirements
- 5-10 years of experience in Data Science, Machine Learning, AI research, and analysis
- Structured thinking and logical organization in writing.
- High tolerance for ambiguity, capable of organizing messy notes into a coherent outline.
- Executive presence, comfortable interviewing senior leaders and providing insightful feedback.
- Ability to work cross-functionally across ML researchers to clients.
Nice to have
- Experience in Model Evaluation, ML Engineering or Technical Research.
- Experience designing or curating datasets (RLHF, SFT data).
Culture & Benefits
- Comprehensive health, dental and vision coverage.
- Retirement benefits.
- Learning and development stipend.
- Generous PTO.
- Potential for a commuter stipend.
- Inclusive and equal opportunity workplace.
Будьте осторожны: если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого - это мошенники. Обязательно жмите "Пожаловаться" или пишите в поддержку. Подробнее в гайде →