Imagery Analyst - AI/ML Training Data

What you'll do

This is bounding-box annotation with a defence client. A model has already drawn boxes around objects in still images from drone and maritime sensors (daylight and thermal), and your job is to make those boxes right:

• Tighten, correct or remove machine-generated bounding boxes
• Find and label the objects the model missed
• Keep every label consistent with the project guidelines
• Submit finished images inside a secure Navy environment
• Write up ambiguous cases and guideline questions clearly

The project is expected to move from stills to video frames as it grows. Correcting a model's boxes is faster than drawing from scratch, but it rewards a particular kind of attention: the errors that matter are the plausible-looking box that is slightly off and the small object nobody flagged.

Who fits

• Based in the US
• Steady attention to detail on repetitive, accuracy-critical work
• Experience with data labelling, bounding boxes or similar data preparation
• Familiarity with aerial, maritime or sensor imagery, especially daylight and thermal formats
• A record in high-volume, quality-controlled annotation (robotics or autonomous systems labelling is a plus)
• Comfort with dedicated labelling software and willingness to learn new tools
• Clear written and verbal communication

The ad lists US citizenship and the ability to obtain a DoD Common Access Card (CAC), or willingness to go through sponsorship, as preferred. Because the work happens inside a secure Navy environment, treat that as close to essential in practice and ask early whether non-citizens can be onboarded at all. No prior AI experience is needed.

What it pays

$20–60/hour, micro1's published band, but the ad also says compensation is based on each image you submit. Read that as a per-image rate that micro1 expresses as an hourly equivalent: your real hourly figure depends on throughput, and the top of a 3x band is likely reserved for fast, accurate annotators. Ask for the per-image rate and the expected images per hour before you commit.

Worth knowing

Good:

• A wide band with a high ceiling for fast, accurate annotators
• Part-time or full-time, by the ad's own description
• Experience in secure, defence-grade annotation is valued elsewhere
• A planned move into video frames, so the project is meant to continue

Less good:

• Pay appears to depend on output, not hours
• US only, and citizenship plus CAC eligibility will likely matter
• Repetitive, volume-driven work
• Only 10 openings
• No weekly hours or duration published
• Contractor engagement: no benefits and your own tax to manage

About this listing

Posted by micro1 on its own jobs portal on 7 October 2026 and read on 12 October 2026. The $20–60/hour band, 10 openings, US location, per-image compensation note and Navy environment are its own published text. See micro1.