Rerun Technologies AB
Rerun builds a unified data layer for Physical AI, covering data collection through training. Its open-source SDK and commercial Hub ingest, query, curate, visualize and stream multi-rate, multimodal physical-world data for robotics, spatial computing and embodied-AI teams.
At a Glance
- Robotics and embodied-AI companies
- Autonomous-vehicle and drone developers
- Spatial-computing and AR teams
- Computer-vision engineering teams
- +2 more
AI Tools by Rerun Technologies AB
(1)
Rerun
Open Source Physical AI Data Layer
Discussions
No discussions yet
Be the first to start a discussion about Rerun Technologies AB
Latest News
Spotlight: Depth beyond the LiDAR — Rerun documented depth-model experiments on ARKitScenes and multimodal data-layer workflows.
Spotlight: 3D reconstruction is a data layer problem — a worked ARKitScenes example using recordings, the viewer, dataframe API and PyTorch dataloader.
A new data layer for robot learning — SDK 0.32 expanded Rerun toward collection-to-training workflows and introduced Rerun Hub.
Rerun 0.27 released with flexible coordinate frames, Python server management, blueprint improvements and improved time controls.
Products & Services
Open-source SDK and viewer for logging, storing, querying, transforming, visualizing and training on multi-rate, multimodal data. It includes Python, Rust and C++ APIs, an interactive desktop/web viewer, SQL and dataframe queries, a visualization framework, PyTorch dataloader, local catalog and columnar .rrd format.
Commercial, managed backend/data catalog for large-scale Physical AI data. It provides persistent cataloging, byte-range indexing, querying across object stores, dataset streaming to GPUs, link sharing, authentication/SSO and single-tenant deployment in a chosen cloud region.
Market Position
Rerun positions itself as an open-source, robotics-native data layer spanning collection, visualization, querying, transformation and training rather than only an experiment tracker or labeling tool. Its starting thesis distinguishes it from adjacent infrastructure such as Scale AI, Weights & Biases and Hugging Face, while its differentiators are a common multimodal data model, synchronized visualization, columnar physical-data storage and a path from local SDK files to managed Hub/object-storage scale.
Leadership
Founders
Nikolaus West
Stanford-trained computer-vision practitioner who studied computer vision with Fei-Fei Li; joined 3D body-scanning company Volumental after graduation and later became Rerun co-founder and CEO. He also co-founded Grapic.
Emil Ernerfeldt
Computing-science graduate and high-performance graphics/software engineer who came from the gaming industry. He created the egui Rust GUI framework and has worked at/with Magic Leap, Zipline and Meta Reality Labs; he is Rerun's co-founder and CTO.
Moritz Schiebold
Former CEO of 3D body-scanning company Volumental and former colleague of Nikolaus West and Emil Ernerfeldt; Stockholm School of Economics alumnus. He is Rerun's co-founder and COO.
Executive Team
Nikolaus West
Co-founder and CEO
Stanford computer-vision alumnus; former Volumental engineer and co-founder of Grapic.
Emil Ernerfeldt
Co-founder and CTO
Computing-science and graphics engineer, creator of egui; background spanning gaming, Magic Leap, Zipline and Meta Reality Labs.
Founding Story
West says he started Rerun after more than a decade seeing computer-vision teams build physical-world products without broadly available tools for rapid data capture and visualization. At Volumental, visualization-driven development and strong internal tooling helped the team ship ahead of competitors; Rerun was started to make that kind of observability and iteration infrastructure broadly available to teams building computer vision, robotics and other physical-world systems.
Business Model
Revenue Model
The Rerun SDK is free/open source. Rerun monetizes Rerun Hub as a managed, single-tenant commercial deployment over the customer's own S3-compatible object storage; Hub is priced per deployment based on team size and data scale under a contract.
Pricing Tiers
Open-source forever; Apache-2.0/MIT; local/self-hosted toolchain with community support via Discord and GitHub.
Commercial managed service; single-tenant, in the customer's chosen region; contract pricing based on team size and data scale.
Target Markets
- Robotics and embodied-AI companies
- Autonomous-vehicle and drone developers
- Spatial-computing and AR teams
- Computer-vision engineering teams
- Physical-AI research labs and universities
- Teams operating large multimodal datasets and model-training pipelines
- Robotics and robot-learning data collection, curation, training and evaluation
- Autonomous vehicles and drones
- Spatial computing and egocentric/AR datasets
- Computer-vision debugging and observability
- Teleoperation, fleet logs, human-data rigs and simulation data
- 3D reconstruction, SLAM and multimodal sensor-fusion workflows
- Hugging Face LeRobot
- Meta Reality Labs / Project Aria
- Google Research / Brush Gaussian-splatting project
- Unitree Robotics