Data Scientist: Optimize Dataset Value & Incentives

Build AI; San Francisco (CA); On-site; USD 120,000 - 170,000; Full time

Benefits offered by this job

Competitive pay

Medical, dental, and vision coverage

Housing subsidy $2k per month near SF

Relocation support to San Francisco or

Daily lunch and dinner in our office

Unlimited compute budget

Unlimited Codex and Claude credits

Travel

Job summary

Build AI in San Francisco (and Shenzhen) is seeking a role focused on optimizing the dataset objective, the scoring function, and the taxonomy to accelerate data collection quality. You will design marketplace mechanics, run experiments, and work with software to implement incentives and payouts in the collection product.

You will ensure the data collection moves the needle while guarding against misuse. Strong candidates thrive on iteration and translating quantitative results into actionable

Qualifications

• Quantitative: optimize an objective function for dataset quality.
• Experience designing and reading experiments (A/B, holdouts).
• Comfort with messy field/collection data.
• Prefer iterative, data-driven development.

Responsibilities

• Treat dataset quality and value function as the primary objective.
• Design marketplace mechanics for high-value data collection.
• Run experiments and select effective treatments while discarding others.
• Measure data collection impact and translate results for the product.
• Collaborate with software to ship incentives and payouts in the collection product.
• Monitor for Goodhart and close incentive holes.

Skills

Objective function optimization

Experiment design

Data analysis

Handling messy field data

Iterative development

Job description

Build AI in San Francisco (and Shenzhen) is seeking a role focused on optimizing the dataset objective, the scoring function, and the taxonomy to accelerate data collection quality. You will design marketplace mechanics, run experiments, and work with software to implement incentives and payouts in the collection product.

You will ensure the data collection moves the needle while guarding against misuse. Strong candidates thrive on iteration and translating quantitative results into actionable