01Data Annotation & Labeling
Image bounding boxesPolygon & semantic segmentationKeypoints & poseVideo object trackingLiDAR & 3D point cloud+4 more
Muster
Real people. Real work. Powering the AI economy.
Paid, flexible work in annotation, AI training, robotics data, review, collection and customer experience — matched to your skills, languages, home setup and the half-hour blocks you choose.
Built for crowds in India, the Philippines, Africa and LatAm
AI Training & RLHFRanking · Prompt writing · Red-teaming · Fact-checking · Code & reasoning
Robotics & Physical AITeleoperation · Task demos · Egocentric video · Fleet monitoring
Data CollectionSpeech · Photo & video · Documents · Field audits · Sensor data
Language & LocalizationTranslation · Transcreation · Linguistic QA · Subtitling
Quality & DeliveryQA auditing · Gold sets · Adjudication · Team lead · Workflow ops
Data Annotation & LabelingBounding boxes · Segmentation · Keypoints · Video tracking · OCR · Audio
Trust, Safety & ReviewContent moderation · Policy review · Search rating · KYC · Fraud
Customer ExperienceVoice · Chat & email · Tech support · Sales · Back-office
Muster
Real people. Real work. Powering the AI economy.
Paid, flexible work in annotation, AI training, robotics data, review, collection and customer experience — matched to your skills, languages, home setup and the half-hour blocks you choose.
Built for crowds in India, the Philippines, Africa and LatAm
Data Annotation
AI Training
Robotics
Customer Experience
Image bounding boxesPolygon & semantic segmentationKeypoints & poseVideo object trackingLiDAR & 3D point cloudText classification & NERDocument & OCR extractionAudio transcription & diarizationMedical imaging annotationPreference ranking (A/B)Rubric-based response ratingPrompt & conversation writing (SFT)Red-teaming & adversarial testingFact-checking & groundingCode generation & reviewMath & STEM reasoningDomain expert reviewAgent & computer-use evaluationMultimodal model evaluationRobot arm teleoperationHumanoid & mobile robot teleoperationEgocentric video captureTask demonstration recordingFleet monitoring & remote assistIntervention & failure labelingSimulation & synthetic data QAContent moderationAds & policy reviewSearch & recommendation rating
Maps & geo verificationE-commerce catalog QAKYC & document verificationFraud & risk reviewSpeech recordingConversational audioPhoto & video captureHandwriting & document samplesField collection & store auditsNative-language text creationWearable & sensor dataInbound voice supportChat & email supportTechnical support (L1/L2)Sales & lead qualificationCollectionsBack-office processingAI agent supervisionTranslationTranscreation & copy adaptationLinguistic QASubtitling & captioningRemote interpretationQA auditing & calibrationGold sets & guideline authoringAdjudication & tie-break reviewTeam lead / pod leadWorkflow & tooling ops
The work
Eight tracks. Fifty-seven skills.
Work that didn’t exist five years ago.
Pick everything you can do and how well — from “learning” to “I train others”. Most projects confirm it with a short paid test.
0102AI Training & RLHF
Preference ranking (A/B)Rubric-based response ratingPrompt & conversation writing (SFT)Red-teaming & adversarial testingFact-checking & grounding+5 more
03Robotics & Physical AI
Robot arm teleoperationHumanoid & mobile robot teleoperationEgocentric video captureTask demonstration recordingFleet monitoring & remote assist+2 more
04Trust, Safety & Review
Content moderationAds & policy reviewSearch & recommendation ratingMaps & geo verificationE-commerce catalog QA+2 more
05Data Collection
Speech recordingConversational audioPhoto & video captureHandwriting & document samplesField collection & store audits+2 more
06Customer Experience
Inbound voice supportChat & email supportTechnical support (L1/L2)Sales & lead qualificationCollections+2 more
07Language & Localization
TranslationTranscreation & copy adaptationLinguistic QASubtitling & captioningRemote interpretation
08Quality & Delivery
QA auditing & calibrationGold sets & guideline authoringAdjudication & tie-break reviewTeam lead / pod leadWorkflow & tooling ops
How it works
From application to first shift, with nothing hidden.
- 01
Build your profile
Skills, languages, the hours you want, your computer and connection. Paste a résumé and we pre-fill it — you approve every line.
About youSkillsAvailabilityWorkstation - 02
Get verified, not guessed
Your connection is measured in the browser and your device is detected. A recruiter reviews you in 2–5 working days.
148Mbps ↓96Mbps ↑9msMeasured in your browser - 03
See every requirement
Projects show each requirement and whether you meet it — and the one change that would make you eligible.
LiDAR & 3D · AdvancedEnglish B2+16 GB RAMPower backupNight window covered - 04
Accept your offer
Rate, hours, shift in your time zone and start date, upfront. Accept or decline in one tap.
New offerLiDAR annotation — night pod$6.50/h · 30 h/week · starts MondayAcceptDecline
Real people. Real work.
Not a black box.
Every person in a Muster crowd has a real profile, a measured setup and hours they chose. Teams see exactly who they’re working with — and workers see exactly what each project needs.
8
skill tracks
57
skills, four levels each
48
languages with CEFR levels
30
minute scheduling blocks
What you need
An honest list, before you start.
Projects set their own minimums and show them upfront. You don’t need everything to begin — many collection projects only need a phone.
Computer
Laptop or desktop · Windows 10+, macOS or Linux
8 GB RAM; 16 GB for 3D and video
Internet
Broadband you can rely on
10 Mbps down / 3 up — measured in-app
Power
A UPS, inverter or generator
Needed for most shift-based projects
Space
A quiet room you can close
For voice, speech and secure work
Phone
A recent Android or iPhone
For collection and on-the-go tasks
You
18+ with the right to work where you live
A self-declaration — never ID numbers in the form
For operations teams
Recruit, qualify and deploy a crowd — from one console.
- Invite at scaleOne person or thousands from a CSV or a paste from Excel, with row-level validation and campaign tracking.
- Explainable matchingHard requirements and a ranked fit score for every project — plus who is one requirement away.
- Gemini where it removes typingDraft job posts from a brief, search the pool in plain English, pre-fill profiles from résumés. It never decides who gets work.
Supply by hour · a typical week
ISTIllustration of the console’s supply heatmap: approved people available by hour, in the time zone your client works in.
Questions
Before you apply
Is it free to join?
Yes. Muster never charges workers anything — no sign-up fee, no training fee, no equipment you have to buy from us.
How much will I earn?
Every project shows its rate before you apply, per hour or per task. Offers repeat the exact rate and hours before you accept.
How does matching work?
Each project lists hard requirements — skills and level, languages, location, hours, and setup like RAM or bandwidth. You see exactly which you meet and which you don’t. No AI decides who gets work.
What happens with my background declaration?
It’s a self-declaration — we never ask for ID numbers in the form. If a project needs checks, a licensed partner contacts you after an offer, only with your consent.
Can I work on more than one project?
Yes, as long as the hours don’t overlap. Your availability is shared across projects so you aren’t double-booked.
Who can join?
Anyone 18 or older with the right to work where they live. Some projects are limited to certain countries, languages or on-site hubs.
Ready when you are.
Set up your profile in about fifteen minutes. Pick up where you left off any time.
Join the crowd