SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Physical Intelligence is building general-purpose AI systems to control robots for any task. The Data Operations team ensures that the massive amounts of multimodal training data sourced from third-party vendors and partners worldwide are high-quality, well-understood, and safe for model training.
In this role, you will own the end-to-end quality assurance and audit process for externally sourced data. You'll build and run sampling strategies, review workflows, pass/fail decision frameworks, and audit trails to verify data before it reaches training pipelines. You'll work closely with researchers who consume the data and the teams who create it.
Key responsibilities include:
- Designing and implementing the complete QA process for multimodal data ingestion from third-party vendors and partners
- Partnering with researchers to translate quality requirements into concrete, measurable rubrics and acceptance criteria
- Building software interfaces, tooling, and infrastructure to scale QA processes with automation
- Implementing internal QA checks, tracking vendor data quality trends, and surfacing insights back to partners
- Managing vendor quality loops by delivering structured feedback, coaching vendors on their own QA implementation, and enforcing standards through review processes
You bring approximately 5 years of experience building customer-facing software, with demonstrated vendor management experience working directly with third-party data providers. You have strong operational acumen with a bias toward instrumentation, measurement, and rapid iteration. You're comfortable being hands-on with data—reviewing samples, writing and refining quality rubrics, and spot-checking program outputs. You communicate clearly across researchers, operations teams, and external partners.
Nice-to-have qualifications include experience with robotics, egocentric video, multimodal sensor data, defining QA/validation tooling requirements, or collaborating with engineers on automated data quality checks.
