Claru

Training-data platform for physical AI: egocentric video, robotics, and world-model datasets. 10,000+ remote collectors in 100+ cities. Geo-tiered pay ~$5 to $15/hr, paid bi-weekly.

AI Training AI Training Data / Physical AI / Robotics and Video Data
Visit Claru
★ ★ ★ ★ ★ New (0 reviews)
Remote-First (Global, 100+ cities across 6 continents) HQ

What is Claru?

Claru is a training-data company that describes itself as 100% focused on physical AI: it builds purpose-built datasets for frontier robotics, embodied AI, and world models. Rather than synthetic or scraped material, it supplies licensed real-world video across domains including egocentric (first-person) footage, game environments, driving, cinematic, manufacturing, cooking, warehouse, and general human activity. Each clip ships enriched: depth maps, pose estimation, segmentation masks, and structured metadata, with expert human labelers adding what machines miss, namely intent, context, and edge cases. The work is powered by a global network of trained data collectors, reported at 10,000+ collectors across 100+ cities on six continents. Claru cites a track record including 500K+ egocentric videos captured for a world-modeling lab, 10K+ hours of game-environment data, millions of video annotations powering RLHF for frontier video generation, and partnerships with 5+ frontier labs. For workers, Claru is a source of fully-remote contractor gigs: egocentric video capture, gameplay capture, annotation, and expert data labeling, recruited through claru.ai/jobs and claru.ai/for-annotators. Pay is reported as a flat geo-tiered rate in USD paid bi-weekly. Note that Claru does not publicly disclose founders, a headquarters, headcount, or funding, so treat corporate details as unknown, and note it is unrelated to Clarum (private-markets diligence) or Clarifai.

Mission & values

Build the real-world video and annotation datasets that physical AI needs, so robots, embodied agents, and world models learn from licensed human experience rather than synthetic or scraped data.

Qualifications

Claru recruits two contributor tracks, both fully remote. (1) Data collectors: people who capture real-world video in their own environment, most notably egocentric (first-person, often head-mounted) footage of everyday activities such as cooking, manufacturing, warehouse work, or driving, plus gameplay capture. This is accessible work; the main requirements are the right recording setup, a qualifying location, and the ability to follow capture protocols precisely so the footage is usable for training. (2) Annotators and expert labelers: people who label video for intent, context, and edge cases, and support RLHF workflows for frontier video and world models. Domain familiarity helps for specialist footage. Sign-up runs through claru.ai/jobs and claru.ai/for-annotators. Availability and rate depend heavily on your country because pay is geo-tiered, and work is project-based rather than a guaranteed schedule, so treat it as flexible supplemental income and confirm current project availability before relying on it.

Hiring process

  1. 1

    Apply at claru.ai/jobs or /for-annotators

    Pick the track that fits: data collection (egocentric or gameplay video capture) or annotation and expert labeling. Applications run through the Claru site.

  2. 2

    Eligibility and setup check

    Confirms your country (pay and availability are geo-tiered), recording setup for capture roles, and relevant domain experience for annotation roles.

    About 5 days

  3. 3

    Onboarding and capture protocol training

    Training on the capture or annotation protocol, since usable data depends on following specs precisely (framing, duration, metadata, labeling standards).

    About 5 days

  4. 4

    Sample submission and review

    Submit initial footage or annotations for quality review against the project spec before being cleared for volume work.

    About 7 days

  5. 5

    Work projects and get paid bi-weekly

    Claim available projects on a flexible schedule. Payment is reported as bi-weekly in USD at your geo-tiered rate.

    About 14 days

Funding

Investors
No publicly-disclosed funding or corporate details

Awards & recognition

  • 10,000+ trained collectors across 100+ cities on 6 continents · 2026

    Claru

  • 500K+ egocentric videos captured for a world-modeling lab · 2026

    Claru

  • 5+ frontier AI lab partnerships · 2026

    Claru

Company information

  • Industry: AI Training Data / Physical AI / Robotics and Video Data
  • Location: Remote-First (Global, 100+ cities across 6 continents)
  • Website: claru.ai ↗
  • LinkedIn: Company page ↗

Frequently asked questions

What is Claru?
A training-data company focused entirely on physical AI. It builds licensed real-world video datasets (egocentric, driving, manufacturing, warehouse, cooking, game environments) enriched with depth maps, pose estimation, segmentation masks, and metadata for robotics, embodied AI, and world models.
What work can I do for Claru?
Two remote tracks: data collection (capturing egocentric first-person video of everyday activities, or gameplay capture) and annotation / expert labeling, including RLHF support for frontier video models.
How much does Claru pay?
Reported flat geo-tiered rates in USD: about $15/hr in the US, $10/hr Canada, $7/hr LATAM, and $5/hr Asia, paid bi-weekly. This is lower than peers like Outlier or DataAnnotation, so weigh it against alternatives for your country.
Do I need experience?
Capture work is accessible and mainly needs the right recording setup, a qualifying location, and careful adherence to the capture protocol. Annotation and expert labeling benefit from domain familiarity.
Is it remote and open worldwide?
Yes, the work is fully remote and the collector network spans 100+ cities across six continents, but your rate and the volume of available projects depend heavily on your country because pay is geo-tiered.
Is Claru the same as Clarum or Clarifai?
No. Claru (claru.ai) is a physical-AI training-data company. It is unrelated to Clarum, which automates private-markets diligence, and to Clarifai, a computer-vision platform.
Who runs Claru?
Not publicly disclosed. The company does not list founders, a headquarters, headcount, or funding publicly, so treat corporate details as unknown and verify terms directly before committing to work.

Stay in the loop.

One email per week, 5 hand-picked roles.