Li Junyi (LJ)
Co-founder
Grew up in Nigeria. Speaks English, Nigerian Pidgin, and Mandarin.
Real-world data from places other datasets don't reach.
runs on-the-ground data collection in Nigeria for robotics and physical-AI teams: egocentric video, collected to your spec, with consent and QC handled locally.
What we do
-
Field collection to your spec.
We use your capture protocol and hardware, or standard head-mounted phone rigs, and recruit contributors who fit the task.
-
Consent and compliance on the ground.
Every contributor is vetted in person and signs a written consent and data-rights release. We follow Nigeria's Data Protection Act 2023.
-
QC and local payments.
We review footage against your spec before delivery and handle contributor payments locally.
Why us
New geography.
West African homes, workshops, factories, and markets are thin in today's datasets.
People platforms don't reach.
Mechanics, artisans, drivers, factory and technical-school workers, many of whom will never sign up on a gig app.
Local operators, not a sign-up link.
We recruit through people we know and check contributors face to face.
Languages.
Our team speaks English, Nigerian Pidgin, and Mandarin.
What we collect
- Egocentric (first-person) video of real work: workshops, trades, factory tasks, household tasks
- Task and environment metadata
- Basic annotations on request (task segments, success/failure labels, language descriptions)
Need something else, such as wrist cameras or a specific task list? Tell us the spec.
How a pilot works
You send a spec: the tasks, hours, format, and acceptance criteria.
We recruit and brief contributors and confirm consent.
We collect and QC against your acceptance criteria.
You review a sample before the full delivery. You pay for accepted hours.
We suggest starting small: about 10 hours from 5 contributors.