The company is building foundational models for general purpose manipulation, using rich multi-modal human demonstration data generated at scale via its data infrastructure. These models are trained to work reliably across tasks, objects, and robot embodiments and are already live in deployment across multiple verticals.
As a data engineer, you will architect the highest quality robotics dataset in the world, owning the pipelines that turn raw sensor data into training-ready datasets, building tooling for data exploration, filtering, versioning, and release management, implementing quality checks to catch bad data before training, and working in a tight feedback loop with the model team to improve data for training.
Responsibilities
- Build pipelines that turn raw multi-modal sensor data into training-ready datasets
- Build tooling for data exploration, filtering, versioning, and release management
- Implement quality checks that catch bad data before training
- Work in a tight feedback loop with the model team to improve data for training
Requirements
- Strong Python and experience building data pipelines at scale
- Experience with multi-modal or time-series sensor data (video, IMU, force - any of these)
- Experience with data validation at scale - anomaly detection, distribution checks, or debugging corrupted data
- Proof of decision-making with incomplete information
Benefits
- Full-time, on-site in Pune
Additional details
- We're building foundational models for general purpose manipulation. Our data infrastructure allows us to generate rich, multi-modal human demonstration data at exponential scale.
- We are using this data to train general-purpose manipulation models that work reliably across tasks, objects, and robot embodiments. We are live in deployment across multiple verticals.
- Our culture is built on honesty and willingness to fail fast. If you see yourself fitting here, please apply.
- We are developing foundational models for general purpose manipulation, trained on human demonstration data collected from our data factories – facilities where human data generators wear our data collection gloves and perform various tasks.
- We use this data to train custom models that generalize over tasks, objects and work on any robot hardware.
Zobhira
