micro1 is collecting biometric facial data from sibling pairs — 30 unedited selfies and 10 video recordings per person, drawn from the past five years — to train and test facial recognition models on genetically similar faces. Both siblings must apply separately and complete their own submission before either can be onboarded.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work actually is

This is a data contribution engagement, not an annotation or review job. You gather 30 selfies and 10 short video recordings of yourself, spanning roughly the last five years, and submit them through micro1's collection flow. The images and clips must be originals: no Instagram or Snapchat filters, no beauty-mode smoothing, no cropping or retouching, no screenshots of screenshots. If you only have filtered copies of a photo, you either find the original file or leave it out. Reviewers check submissions and send them back with feedback when a file fails — wrong aspect ratio, heavy compression, a face partly out of frame, a filter you forgot was on by default in your phone's camera app.

The sibling requirement is the point of the project. Facial recognition systems are stress-tested hardest by faces that are similar but not identical, and sibling pairs supply exactly that. Practically, this means your onboarding is coupled to someone else's: both siblings submit separate applications through the same process, and neither is onboarded until both are complete. If your sibling stalls, you stall.

What the screen looks for

That you actually have the archive. The single most common failure is applying, then discovering you have twelve usable photos and no video from 2021. Count before you apply.
That your files are unedited and you understand what that means. Candidates who cannot distinguish an original camera file from an exported, filtered copy get screened out.
That your sibling is real, willing, and ready to apply. Expect direct questions about this.
That you understand and consent to biometric use. Your face becomes training data. The screen wants a candidate who has thought about that, not one who skimmed past it.

Logistics and pay

Fully remote and asynchronous. The task itself is short — most people spend an hour or two locating files, checking them against the spec, and uploading — with a possible second round if reviewers request replacements. The posted band of $139–140/hr is as observed on the platform; for a task of this shape it typically resolves to a fixed payout for a completed, accepted submission rather than open-ended hourly work, and it is not guaranteed. Treat it as a one-off engagement, not ongoing income. A $100 referral bonus is advertised separately.

AI Facial Data Collection Associate (Siblings)

What the work actually is

What the screen looks for

Logistics and pay

Interview prep — 10 questions