Skip to main content
C

Calder

Calder builds data infrastructure for embodied AI and world models, capturing and curating high-fidelity real-world spatial data via teleoperation, mobile collection tooling, and automated validation pipelines.

Calder builds the data infrastructure used to train embodied AI and world models. The core work is capturing, processing, curating, and delivering high-fidelity real-world spatial data - 4D and 3D captures, egocentric video, and teleoperated robot demonstrations - intended for model training in manipulation, navigation, and interaction. The datasets are consumed by robotics teams and AI laboratories rather than end users.

On the collection side, Calder operates hands-on capture and quality-control operations built around teleoperation, hardware calibration, and data validation. Automated evaluation pipelines run alongside these steps as part of routine collection and QC, so calibration and validation issues surface during capture rather than downstream in training. Collection and quality-control operations run in Singapore and Malaysia.

Contributor-facing tooling is a large part of the system. Calder's mobile data-collection apps handle task discovery, device provisioning, data capture, uploads, feedback, and payments, and connect directly to collection hardware. This supports a worldwide contributor network doing paid data-collection tasks, with feedback loops wired into the same tooling that handles ingestion.

Engineering work spans the hardware/software boundary: 3D capture and spatial data pipelines, computer vision, data curation, teleoperation systems, and mobile collection tooling. Roles touch embodied AI, world models, and the data engineering required to make real-world spatial data usable at training scale.

1 Open job

We use cookies

We use cookies to understand how this board is used and to improve it. Analytics run only if you accept. Cookie Policy