Example project range
$15-$45/hr Replace with approved ratesNow accepting contributors
Get paid to train the next generation of AI models.
Paid AI training projects for experts and skilled labelers. Remote-first work, clear guidelines, and rates shown before eligible work begins.
$15-$45/hr example base range
Weekly example payout cadence
Remote where projects allow
Profile matched to open work
Example payout cadence
Weekly Confirm before publishingProject dependent
Remote Eligibility varies by locationMatched workflows
Skills Profile and verification may applyPlatform context from teams building and evaluating AI systems
AnthropicNVIDIAHugging FaceDatabricksRocheLockheed MartinU.S. Air Force
Find work that matches you
Different kinds of work. One profile.
Create one profile for AI training, labeling, language, and expert review opportunities. Access depends on availability, requirements, and onboarding fit.
AI response evaluator
Compare model responses. Score for accuracy, helpfulness, safety, and tone. Write short justifications.
Sample - pairwise preference
Response A Lists steps without explaining tradeoffs.
Response B - chosen Names tradeoffs and flags one safety risk first.
$22-$45/hr example
writingjudgment
Data collection contributor
Capture, upload, and structure text, image, audio, video, or sensor data for project-specific AI training needs.
Sample - data collection batch
Audio 24 clips accepted
Images 112 frames uploaded
Metadata location and consent checked
$15-$35/hr example
capturequality
Domain expert reviewer
Bring subject-matter knowledge in medicine, law, finance, engineering, research, and other expert domains.
Sample - expert review
Cite Source is present and relevant. Flag Recommendation needs expert review. Pass Safety note is appropriate.
$60-$180/hr hypothesis
credentialsverified
How it works
How experts improve AI models.
Contributors turn project guidelines into the reliable human feedback AI teams need to ship safer, more useful systems.
Step 01
Review AI outputs
Score responses for accuracy, helpfulness, and safety against a project rubric.Step 02
Label complex data
Apply instructions to image, video, audio, and text data: boxes, transcripts, intents, and more.Step 03
Write expert demonstrations
Compose ideal responses for hard prompts and teach models what good looks like in your domain.Why Workforce
Paid project work, with a quality bar that means something.
Workforce should feel like a curated contributor pool, not a race-to-the-bottom task marketplace. Contributors get matched, projects get reviewed, and expectations are visible up front.
Real projects from AI teams
Each project can list its sponsor, rubric, requirements, and rate before a contributor accepts.
Rates set up front
Contributors see hourly or per-task expectations before they start work.
Built around Label Studio
The same platform teams already use to manage labeling and review workflows anchors the experience.
app.humansignal.com/project/342/queue
Queue - LLM-EVAL-1284
421 / 1,284 - 62%#0421
Clinical safety review
$38.00active#0420
Drug interaction grading
$38.00done#0419
Differential diagnosis
$45.00done#0418
Imaging caption check
$32.00reviewCompensation
Rates are public. So is what changes them.
Each project should list its pay rate up front. Specialty domains, verified credentials, and consistent quality history can move contributors up the bands.
Draft
Ranges below are placeholders and need approved compensation details before publishing.
RoleBase ratePaid as
AI response evaluator$22-$45 / hrper accepted task
Data collection contributor$15-$35 / hrper accepted task
Domain expert reviewer$60-$180 / hrhourly or project-based
Contributor stories
Designed for real human judgment.
Placeholder testimonials for design review. Replace with verified contributor quotes and approved metrics before launch.
The work feels close to research: read carefully, compare outputs, and explain why one answer is stronger.
Maya R.
AI response evaluator
Avg rate$36.50/hrTasks accepted2,184IAA0.94
I can apply technical reasoning to practical AI tasks without needing to join a full-time research team.
Arjun S.
Technical reviewer
Avg rate$112/hrProjects9VerifiedExpert
Active contributors across example regions
Placeholder proofExample annual contributor payout story
Replace before launchGitHub stars on Label Studio
Public proof pointExample median agreement across active projects
Placeholder metric
SOC 2 Type II
GDPR compliant
HIPAA-ready environments
Label Studio ecosystem
Featured proof placeholders
How to apply
From signup to first eligible task.
Most contributors can complete the profile step quickly. Project access depends on active demand.
Create your account
Email and password. Verification starts on the way in.
~2 minBuild your profile
Languages, domains, skills, availability, and preferred work types.
~10 minPass project checks
Some projects require a skill screen, calibration task, or credential review.
same day to 2 daysStart earning
Accept eligible work once requirements, rates, and project details are clear.
paid cadence variesFAQ
What to know before you apply.
For launch, replace placeholder rate and verification language with approved compensation, legal, and operations details.
Who should apply?
People with careful judgment, strong communication, and the ability to follow detailed guidelines. Domain experts can qualify for specialty projects.
How are projects paid?
Each project should show its rate before work starts. The ranges on this design are placeholders until approved compensation details are ready.
How much time do I need?
Time commitment depends on active project needs and your eligibility. Completing a profile helps match available work to your schedule.
Is the work remote?
HumanSignal Workforce is designed for remote contribution where project requirements allow. Some projects may have location or workspace requirements.
Do I need AI experience?
AI experience can help, but many projects rely more on careful reading, domain knowledge, language fluency, or consistent guideline following.
Will I need verification?
Some projects may require identity checks, skill reviews, or credential verification before work begins.
Open right now
Projects accepting contributors today.
Apply once. Rates, requirements, and project details should show before a contributor accepts anything.
Create contributor account
~2 minNo card requiredPaid cadence varies
Workforce - queue
12 projects - 487 seatsClinical reasoning - safety review
CLIN-2104 - 18 seats - remote
Spanish chat evaluation
EVAL-1781 - 42 seats - remote
Python code review
CODE-0894 - 9 seats - remote
Image bounding boxes
CV-3340 - 120 seats - remote
Legal contract redaction
LEG-0612 - 6 seats - remote