Skip to content
All roles
Product

Agent Data Strategy — Product

Define creative-occupation task scenarios with creative experts; design evals, benchmarks and rubrics; turn real creative work into data products.You understand agent products, model capability limits and evaluation methods, and can go deep into real creative workflows to break down complex tasks with experts. Your output is task definitions, expert workflows, task specs and deliverable product plans.

What you’ll do

  1. 01Define task scenarios for creative occupations, working with experts in graphic design, 3D, games, visual interaction, music, and film to decompose expert judgment and real workflows.
  2. 02Design agent evals and benchmarks — task design, metrics, rubrics, success criteria, and quality acceptance.
  3. 03Iterate task design, data standards, eval plans, and product processes based on model performance, expert feedback, customer needs, and delivery results.
  4. 04Use AI tools deeply for research, task decomposition, prototyping, data analysis, documentation, and workflow automation.
  5. 05Work with algorithm, engineering, and delivery teams to ship product plans into runnable task environments, eval systems, data pipelines, and customer deliverables.

What we look for

  1. 01Bachelor’s or above, any major: design, fine arts, animation, game design, music, film, AI, CS, industrial design, etc.
  2. 02AI-product or agent-product experience at big tech or a leading AI company — agent eval, benchmarks, model eval, complex-task assessment, AI product testing, or expert-labeling processes.
  3. 03Ongoing interest in agents, LLMs, post-training, eval, RL environments, and AI data pipelines.
  4. 04Strong task-decomposition ability — turning fuzzy business problems into clear processes, task units, judgment criteria, and acceptance rules.
  5. 05Experience working with creative professionals, customers, or cross-functional teams, quickly grasping professional context and translating it into product language.
  6. 06Strong documentation — data plans, task specs, eval standards, expert guides, and customer deliverables.
  7. 07Comfortable with early-startup ways of working — ambiguity, self-direction, coordinating resources, and driving results.

Nice to have

  • High-frequency AI-tool user, actively using AI to raise research, writing, analysis, prototyping, and delivery efficiency.
  • Able to use AI tools or simple scripts for prototypes, data processing, workflow building, and efficiency tools.
  • A background or portfolio in a creative field, or your own views on what good creative work looks like and how AI should learn it.
Send your CV

Send your résumé and a short note on why this role.