Staff Product Manager, AI Research & Engineering

Collectors•San Francisco, CA

4d•$165,848 - $269,259•Remote

About The Position

This team, AI Research & Engineering (AIRE), builds the applied machine learning systems that run our business at scale, from prototype to production. This role is ideal for a product manager who is equally comfortable sourcing training data and writing evals for our models, standing up production-ready pipelines, and shipping tools that help internal teams make better decisions. You'll work across computer vision, structured data, and agentic AI workflows, with direct impact on products used by graders, researchers, and collectors. Every one of those products is only as good as the data behind it, and you'll own that data from first capture to final correction. As a Staff Product Manager, you'll decide what training data we collect, in what order, and how human corrections flow back into our models, partnering closely with ML Engineering and subject matter experts to improve model accuracy, development speed, and cost across the portfolio. You'll bring the technical fluency to earn credibility with engineers, the judgment to turn ambiguous problems into clear strategy, and the systems thinking to build feedback loops that make every model better than the last.

Requirements

10+ years of product management experience, including experience building a product, platform, or capability from the ground up.
Strategic and execution-oriented: You can set direction, make tradeoffs, influence operational decisions, and drive execution in ambiguous environments.
Experience with AI/ML or AI-Enabled Operational Workflows: You’ve worked on products where both models and human expertise play a role, and understand how to design workflows that leverage both effectively.
Strong systems thinker: You can identify constraints across complex, multi-step workflows and understand how changes in one part of a system impact throughput, quality, and cost elsewhere.
Comfortable owning business outcomes: You’ve owned metrics tied to cost, efficiency, quality, or other measurable business outcomes and use data to drive decisions.
Strong partner and influencer: You build credibility with technical teams, operators, and subject matter experts and can influence effectively without direct authority.
Hands-on builder: You’re comfortable rolling up your sleeves to establish new processes, pipelines, documentation, and operating mechanisms. The first year will involve significant 0-to-1 work.
Exceptional written communication: You can clearly articulate strategy, plans, tradeoffs, and decisions for both technical and business audiences.
Willingness to travel: This role will require travel to multiple data collection sites, approximately once per month to once per quarter.

Responsibilities

Define the strategy for our training data capabilities, prioritize what data we collect and when, and own decisions around quality, review depth, confidence thresholds, and cost.
Own data collection end to end, including procurement, preparation, capture, expert labeling, and annotation, and build scalable systems across each stage.
Coordinate work across teams and sites to maximize throughput, identify bottlenecks, and drive capacity planning across specialties and locations.
Partner with subject matter experts and power users to define review processes and feedback loops that turn human corrections into valuable training data.
Work closely with ML Engineering, Operations, and subject matter experts to set direction, align priorities, and drive execution without relying on direct authority.
Define the metrics that demonstrate data quality, model improvement, throughput, and cost efficiency, and communicate progress and tradeoffs to senior leadership.
Mentor other Product Managers, contribute to product practices, and build documentation and mechanisms that help scale product thinking across the organization.
Use emerging AI tools and coding agents, including Claude Cowork, Claude Code, and Codex, to accelerate product development, improve workflows, and find new ways to operate at scale.