About this role
You’ll own delivery of Taste’s data projects - building the bridge between creative judgment and model training. You’ll work directly with our research, engineering, and growth teams to design data pipelines for subjective domains (design, writing, tone, emotion), coordinate expert Taste-Makers, and partner with AI labs to deliver datasets that actually move model quality.
This role is ideal for someone who understands how training data works, cares deeply about quality and taste, and wants to set the standard for what “great” looks like in AI systems.
What You'll Do
This is a hybrid role (strategy, project management, ops and product)
Partner with research and engineering to design and run data pipelines - defining formats, labeling criteria, QA loops, and evaluation workflows.
Coordinate with Growth to recruit and manage expert Taste-Makers, ensure alignment on timelines and budgets.
Set and enforce the quality bar - define rubrics, run spot checks, and use judgment and taste to decide what’s actually great.
Interface directly with AI labs and customers - manage communication, feedback, and fast iteration.
Identify gaps in tooling and process; work with engineering to build the systems that make these projects scale.
Example Projects
Deliver a UI-design reward model dataset with a frontier AI lab.
Build a writing-tone evaluation pipeline (pairwise + rubric) to benchmark model expressiveness.
Design and run multi-stage QA for a large-scale model personality dataset.
About Taste
Taste is building the taste layer for AI - teaching models what great feels like across subjective domains.The next frontier of intelligence isn’t accuracy - it’s taste. How something sounds, looks, or resonates. We lead research with creative experts and AI labs to develop the methods, data, and evaluations that teach models to understand quality, style, and emotion.If you want a world with less AI slop and more taste, join us.