Data Acquisition Program Manager - Frontier Research Lab

Data Acquisition Program Manager

Palo Alto - 3 days per week

Up to $250k base plus Bonus & Equity

Our client is a frontier AI lab building real-time world models: AI that simulates reality rather than generating clips of it, with the highest physics score among evaluated world models. Their models learn from physical-world data, including robotics demonstrations, egocentric and interaction video, driving data and simulated environments, and they need to acquire a great deal more of it.

The role

This is a senior individual contributor seat and one of the lab's first dedicated data hires. You'll own how physical-world data gets into the building: deciding with researchers what's needed, sourcing it from data providers and collection partners or running the collection yourself, negotiating the terms, and making sure what arrives is good enough to train on. To be clear about what this role is: it's about acquiring and collecting data, not managing annotation of data the company already has. If your experience is mostly labeling programs for existing platform content, this probably isn't the right fit.

What you'll do

• Work directly with researchers across several model teams to turn loose data needs into concrete specs: modalities, scenarios, volumes, quality bars and timelines
• Source and qualify data providers and collection partners across robotics, video and simulation data, and decide which to engage
• Negotiate terms and SOWs, including pricing models, acceptance criteria, delivery schedules and the flexibility to scale volume quickly
• Run or oversee physical data collection programs: protocols, capture setups, operator teams, field logistics and consent
• Build the quality gates that reject unusable data before it reaches training, and track throughput, yield and cost
• Forecast data needs ahead of training cycles and make the trade-offs visible to leadership

What you bring

• Hands-on experience acquiring or collecting physical-world data for model training: robotics, autonomous vehicles, egocentric or teleoperation video, motion capture, or game and simulation environments
• Around five or more years in data programs, with at least one program you owned end to end, from research requirements through collection or purchase to delivery
• Direct experience negotiating with data providers, collection vendors or service partners: pricing, terms, and performance
• Working understanding of how data needs differ across pre-training, fine-tuning and evaluation
• Comfort with data at a technical level: you can query it, check it yourself and use AI tools to speed up your own work

Strong pluses

• Experience with foundation model training data
• Collection or acquisition roles at robotics, AV, physical AI or world-model companies, or at data providers serving them
• Having set up a data function or collection program from scratch

This is a chance to shape how one of the most exciting AI labs in the world gets the data that teaches its models how the physical world works.