Data Partnerships

ABOUT GENERALIST

At Generalist, we are on a mission to build general intelligence for the physical world and make it useful to everyone. We believe the industries and homes of the future will depend on humans and machines working together in new ways. Robots can help us build more and get more done.

We build embodied foundation models, starting with a focus on dexterity. This requires advancing the frontiers of data, models, and hardware, to enable robots to intelligently interact with the physical world.

The company embraces both large-scale AI and robotics as core to its DNA. Our team of researchers, roboticists, and company builders come from OpenAI, Boston Dynamics, Google DeepMind, and other frontier labs—with a track record of shipping AI breakthroughs. Before Generalist, we pioneered large embodied multimodal models and vision-language-action models (PaLM-E, https://research.google/blog/palm-e-an-embodied-multimodal-language-model/ RT-2 https://deepmind.google/blog/rt-2-new-model-translates-vision-and-language-into-action/, Gemini Robotics https://deepmind.google/models/gemini-robotics/), launched and scaled ChatGPT https://chatgpt.com/ and GPT-4 https://openai.com/index/gpt-4-research/ to hundreds of millions of users, engineered the foundations of autonomous driving, built next-generation robots (Atlas https://bostondynamics.com/atlas/, Spot https://bostondynamics.com/products/spot/, Stretch https://bostondynamics.com/products/stretch/) and pushed the limits of what they can do (from parkour https://www.youtube.com/watch?v=tF4DML7FIWk to manipulation https://bostondynamics.com/blog/large-behavior-models-atlas-find-new-footing/, and testing robustness https://www.youtube.com/watch?v=aFuA50H9uek).

We are an equal opportunity employer, and we do not discriminate on the basis of race, religion, color, national origin, sex, sexual orientation, age, veteran status, disability, genetic information, or other applicable legally protected characteristic.

ABOUT THE ROLE

Embodied foundation models and their capabilities are driven by data — and robotics data does not exist on the internet waiting to be scraped. It has to be manufactured by people doing real work in real environments, to a spec guided by our model science and scaling laws.

You will own the data engine that feeds our models: finding the right partners who can collect at quality and volume, standing them up, supplying them with necessary hardware, and holding them to a bar. Our data partnerships and operations extend globally, and expanding the ecosystem involves building the underlying network infrastructure for data ingestion and distribution that will feed in to support future commercial partnerships as well. Expect to extend the data engine into new geographies and new data modalities on tight timelines.

The hardest part of the job is translating model needs and data requirements into something a data partner can execute without you in the room (and you are the first to catch the drift when what comes back is technically compliant but practically useless).

YOU’LL BE RESPONSIBLE FOR:

- Data supply strategy. Identify what we need to buy versus build. Map the vendor landscape across data types and quality. Bring recommendations, not options.

- Vendor sourcing and diligence. Find, evaluate, and pilot new data partners. Run structured trials before committing volume. Know how to tell a real capability from a good deck.

- Requirements elicitation. Sit with ML and engineering leads and extract the actual requirement — volume, environments, embodiments, task diversity, annotation schema, acceptance criteria, what happens downstream. Most of the time the requirement does not exist in writing until you write it.

- Translation and specification. Author partner-facing specs that a non-ML operator can execute against with no follow-up call. Define the unit of delivery, what passes, what fails, and what to do when it is ambiguous.

- Commercial terms. Structure pricing, rate cards, minimums, milestones, and acceptance language. Work with legal on MSAs and SOWs. Own the unit economics and know what we are paying per unit of usable data, not per unit delivered.

- Scaling a partnership into a network. When one partner hits capacity, stand up the next without dropping quality or blowing up cost.

- Operating cadence. Trackers, weekly partner reviews, forecasts against the research roadmap, and a clear picture of cost, volume, and quality that anyone at the company can read.

- Closing the loop. Report back to research on what the data actually produced and use it to change the next cycle's spec.

- Travel. Periodic domestic and international travel to partner sites and collection operations.

YOU MIGHT THRIVE IN THIS ROLE IF YOU:

- 6+ years owning external partners or vendors who delivered a product or service to you — supply chain, vendor management, outsourcing/BPO management, data operations, or partnerships with a delivery obligation.

- Demonstrated experience standing up a new vendor from zero: selection, diligence, contracting, ramp, and ongoing management.

- A track record of translating technical or specialist requirements into unambiguous instructions that a non-technical execution partner delivered against successfully.

- Fluency with commercial mechanics — pricing structures, SOWs, acceptance criteria, unit economics — and the judgment to know which terms actually protect quality.

- Exceptional written communication. You will write documents that people you have never met execute without you present.

- Comfort operating with incomplete requirements and re-scoping mid-flight without losing the relationship.

- Even-keeled under pressure and relentlessly consistent on follow-through.

PREFERRED QUALIFICATIONS

- Experience buying or managing data collection, annotation, or labeling at scale — managed workforce, crowd, or BPO.

- Experience close to an ML or research organization, and enough working understanding of training data to push back on a request rather than just relay it.

- Background in autonomous vehicles, robotics, hardware, or another domain where physical-world data acquisition is a first-class problem.

- Experience at a company that grew quickly enough that the process you inherited stopped working and you had to rebuild it.