Encord

Encord is a multimodal data infrastructure platform for physical AI and enterprise teams, covering the data pipeline from pre- and post-training through deployment. It addresses the need to manage, curate, and annotate data across video, LiDAR, audio, text, images, documents, and sensor data, while keeping label lineage and quality controls in place for production use.

Core capabilities include native video annotation with sensor fusion in a single workflow, embedding-based search and model-in-the-loop curation for finding edge cases, and model alignment through RLHF, rubric-based evaluation, and pairwise comparison. The platform offers an API and SDK, data agents, model evaluation and analytics, active learning pipelines, and integrations with cloud storage including AWS S3, Azure Data Lake Storage, and Google Cloud Storage. Deployment options include cloud, private VPC, and on-premises, with SSO and role-based access controls.

Encord is packaged in three editions: Starter for individuals and small teams prototyping applications, Team for teams scaling a few AI applications, and Enterprise for companies shipping multiple applications across teams. Data volume limits and supported data types vary by plan, with some modalities such as geospatial, ECG, 3D/LiDAR, and DICOM available as add-ons.