Data Annotator - Video Content

$10–12/hrrate as listed

Label and quality-check video footage for AI training sets, tagging objects, actions, and events frame by frame against a written spec. Work is remote and task-based through micro1, with volume that varies week to week.

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

The work

Video annotation is close, repetitive attention paid over long stretches. On a typical shift you open a queue of clips — anywhere from a few seconds to several minutes — and apply the labels a spec calls for: bounding boxes or polygons around objects, action segments with start and end timestamps, event tags, occlusion and truncation flags, and object IDs held consistent across frames so a tracked person or vehicle keeps the same identity when they leave and re-enter the frame.

Most projects run in a browser tool (CVAT, Labelbox, SuperAnnotate, or a client's internal platform). You will spend real time in the guidelines document, because the hard part of this job is rarely the drawing — it is deciding what the spec means when footage is dark, motion-blurred, half-occluded, or shows something the guidelines never anticipated. Strong annotators flag ambiguity instead of guessing silently. Some queues are review rather than annotation: you audit another person's work, accept, correct, or reject with a reason.

What the screen looks for

Evidence you have done this before. Which tools, which label types, roughly what throughput, and what the QA feedback loop looked like.
Spec literalism. Interviewers probe whether you follow written rules over instinct, and whether you can restate an edge-case rule accurately.
Judgment on ambiguity. How you decide between a best-guess label, a flag, and a skip — and how you keep decisions consistent across a long queue.
Sustained accuracy. Willingness to work under sampled audits and hit an agreement threshold, not just move fast.

Logistics

Fully remote and asynchronous, with deadlines set per batch rather than fixed hours. Observed pay for this listing sits at $10–12/hr; rates and available volume are set by the client project and are not guaranteed. Expect a paid or unpaid qualification task before production work, a reliable computer with a real mouse or trackpad and stable bandwidth for video streaming, and periods where a project pauses between batches.

What they look for

Visual Discrimination
Process Adherence
Written Clarity
Attention to Detail

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Have you used at least one professional video annotation platform (such as CVAT, Labelbox, SuperAnnotate, or a custom annotation tool) to label video footage?
Video annotation tool experience · required
Have you completed a video labeling project where your work was audited against a written specification and you received quality feedback (agreement score, rejection rate, or corrective feedback)?
Spec-based labeling · required
Do you have access to a computer with a real mouse (not trackpad-only) and stable internet capable of streaming video without frequent buffering?
Reliable computer and internet · required
Can you reliably meet per-batch deadlines without fixed hours or daily check-ins, and do you have the ability to work across multiple time zones or at variable times?
Asynchronous work availability · preferred
Have you labeled videos where you maintained object identity (the same ID for a person or vehicle across frames, including when occluded or off-screen)?
Object tracking or multi-frame labeling · preferred

0 of 5 answered

Included with this listing

Interview prep — 11 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Whether you have genuine hands-on experience with the core task, not just exposure to the concept—and whether you can articulate concrete metrics that show you know how quality is measured in this work.
A strong answer: Name a real tool (CVAT, Labelbox, SuperAnnotate, or a platform), describe a specific label type you applied (bounding boxes, polygons, action segments, object IDs across frames), and cite a real QA metric—agreement percentage, rejection rate, or feedback you received—that proves you've been through an audit cycle.
Where people lose it: Claiming 'video labeling experience' without naming a tool or label type, or giving vague throughput numbers ('hundreds of clips') with no mention of how your work was actually graded—signals you may have done something tangentially related but not the job itself.