Staff Technical Product Manager, Embeddings & Search

About the role

You set the strategy and roadmap for Marengo and Search. You work with the research team on what the model should learn, how to evaluate it, and when it is ready to ship. You work with customers and field engineers to understand where retrieval breaks in production and what they will need six months from now.

Your week splits roughly three ways: research partnership, customer and field work, and internal product execution. The scope is the full stack: evaluation data definitions, model evaluation, release cadence and management, ranking quality, the search API, and deployment across managed SaaS, customer hosted environments, and AWS Bedrock.

Responsibilities

• Set the product strategy and roadmap for Marengo and Search, deciding what gets built, what gets deferred, and what gets killed
• Partner with the Marengo research team on model quality: eval rubrics, training data investments, release readiness
• Partner with the GTM on launch planning, execution, and enablement including post launch monitoring
• Spend real time with customers and field teams understanding where retrieval fails in production and anticipating what they will need next
• Define the quality bar for retrieval and hold it across every release and every deployment
• Own how embeddings and search get deployed across managed SaaS, customer hosted environments, and AWS Bedrock
• Stay sharp on the competitive landscape

Qualifications

• You have a research, ML, or engineering background with real work in retrieval, embeddings, vector search, or multimodal models, and you moved toward product because you care more about what gets built and why
• You have been a senior solutions engineer or forward deployed engineer with deep ML understanding, and you have been the de facto product owner on the hardest customer problems whether or not the title was yours
• You can go deep on retrieval architecture tradeoffs with a researcher in the morning and frame a product decision for a GTM team in the afternoon, and both conversations are substantive
• You have strong opinions about what makes search work in production and can back them with evidence, not intuition
• You have strong opinions on how to best serve humans and agents as distinct customer segments
• You see what customers need today and can extrapolate what they will need next. You use current demand as a foundation for roadmap decisions, not just a backlog

Preferred Qualifications

• 5 to 8 years of experience, though what matters is demonstrated capability, not tenure
• 3+ years of shipping products with a model related core
• Experience with multimodal models and the operational cost of running them at scale
• Experience in video language models
• Experience augment product development and releases with modern AI tooling