The role in one line
You photograph yourself following project specifications to build diverse facial datasets for AI training systems.
Written by Training Turk from the public listing; it may be incomplete or out of date. Read the full posting on micro1.
What you would do
- Follow detailed instructions for posing, head angle, facial expression, and framing for each photo
- Control your environment to achieve consistent, adequate lighting without shadows or glare on your face
- Review each image for sharpness, proper exposure, and adherence to pose requirements before submitting
- Complete demographic questionnaires paired with your photos, including details about appearance and background
- Retake photos as many times as necessary to match project specifications exactly
Who they are looking for
- Age 18 or older, based in North America, Latin America, Europe, Middle East, or Africa
- Owns a smartphone capable of taking clear, well-exposed self-portrait photographs
- Comfortable with self-photography and able to follow detailed visual instructions precisely
- Able to assess photo quality independently, recognizing focus, lighting, and pose issues
- Available to retake photos multiple times until meeting exact project requirements
Skills this role asks for
What the interview is likely to probe
1.Assessing image technical quality
AI facial datasets require sharp, well-lit images. Submitting blurry or poorly exposed photos wastes time and won't be approved. You need objective judgment.
Expect something like: “You've taken 4 photos in your current location. One is slightly out of focus, one has harsh shadows on one side of your face, and two look clear. Which do you submit?”
2.Matching precise pose angles
Facial recognition training requires specific angles - sometimes 3/4 profile, sometimes straight-on, sometimes tilted. You must be able to reproduce these exactly.
Expect something like: “The project specifies a 45-degree angle looking toward the camera with a neutral expression. How would you ensure your head position is exactly 45 degrees and not 40 or 50?”
3.Recognizing and fixing lighting problems
Poor lighting creates unusable training data. You must spot issues like window glare, shadows under eyes, or overexposure and adjust your setup.
Expect something like: “You're photographing near a window at 3pm. The sunlight creates harsh shadows across one side of your face. What quick adjustments could help?”
4.Completing demographic questionnaires accurately
Your honest responses tag the dataset so models learn from properly labeled examples. Inconsistencies between photos and questionnaire data corrupt training sets.
Expect something like: “A question asks about your ethnic background with multiple options. How do you answer if your background is mixed, and what if your photo appears ambiguous?”
5.Persistence with retakes
Getting perfect pose and lighting alignment often requires 5-10 attempts. You must have patience and precision to meet exact specifications rather than submitting close-enough work.
Expect something like: “You've retaken a photo 8 times and it's still slightly off-angle. You've been working on this for 20 minutes. How do you approach the 9th attempt?”
Exercise you may get
Take a self-portrait following three specific requirements: 3/4 angle to camera, natural lighting without shadows, neutral expression. Evaluate whether your photo meets these specs and explain what you'd adjust if retaking it.
How to prepare
- Test your smartphone's camera quality in various lighting conditions to understand your device's capabilities and limits
- Practice taking self-portraits in consistent poses and angles, using mirrors or reference images to verify alignment
- Learn to identify common photo quality issues like focus softness, overexposure, underexposure, and directional shadows
- Research how facial recognition datasets are used so you understand why consistency and quality matter for the project
Facts
- Eligible locations
- Bangladesh, Hong Kong, India, Indonesia, Japan, Kazakhstan, Kyrgyzstan, Malaysia, Pakistan, Philippines, Singapore, Sri Lanka, Taiwan, Thailand, Uzbekistan, Vietnam, Austria, Belarus, Belgium, Denmark, France, Germany, Greece, Italy, Netherlands, Portugal, Russia, Spain, Switzerland, United Kingdom, Argentina, Brazil, Chile, Colombia, Mexico, Peru, Algeria, Bahrain, Egypt, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Palestine, Qatar, Saudi Arabia, Tunisia, United Arab Emirates, United States, Canada, Nigeria, Kenya, South Africa, Ghana, Ethiopia
- Domain
- Robotics
- Role type
- expert
- Posted
- 9/17/2026
- Open slots
- 3000