Skip to content
AI Trainer Jobs
Companies

Project Beid - Korean E-commerce Product Matching Annotation Reviewer

Welo Data · 100% remote · Contract · Posted

Pay
$27.60/hr
Location
United States
Languages
Korean, English
Hours
Flexible
Openings
Not listed
Level
Experienced

Summary

Review Korean e-commerce product matching annotation decisions, check high-risk rows, correct errors, give annotator feedback, and validate deliverable compliance. Requires native Korean proficiency.

How the pay compares

This job pays $28 an hour.

  • Languages jobs: median $36 an hour across 515 open hourly jobs. This job pays $8 less.
  • Welo Data jobs: median $25 an hour across 54 open hourly jobs. This job pays $3 more.

What you'll do

  • Review annotators' ACCEPT/REJECT decisions and reject reason codes
  • Sample-check or fully review high-risk rows
  • Correct errors and give feedback to annotators
  • Compile client questions and maintain a clarification log
  • Validate structural compliance of deliverables
  • Track and report accuracy and unresolved rows
  • Monitor progress to ensure on-time delivery

Requirements

  • Native proficiency in Korean
  • 1+ years of experience in data annotation, content moderation, or QA
  • Strong judgment on edge cases with clear explanations
  • Clear communication and constructive feedback skills
  • Team lead or training experience preferred

Skills

  • Data Annotation
  • Quality Assurance
  • Korean
  • Content Moderation
  • Data Analysis
  • Communication

Full description

Responsibilities 

- Review annotators’ ACCEPT/REJECT decisions and reject reason codes, with a focus on preventing false matches and false rejects. 

- Sample-check or fully review high-risk rows (multiple codes, OTHERS, rows with uncertainty notes). 

- Correct errors, give feedback to annotators, and identify recurring error patterns to keep rule interpretation consistent. 

- Compile questions for the client where rules are ambiguous, and maintain a clarification log. 

- Validate structural compliance of deliverables: every input row returned exactly once and unchanged, timestamps, exact decision and code spelling, correct separators, valid CSV, identifiers preserved as text. 

- Track and report accuracy, review coverage, known limitations, and unresolved rows to support the summary report. 

- Monitor progress to ensure on-time delivery. 

 

Requirements 

- At native proficiency level of Korean. 

- 1+ years of experience in data annotation, content moderation, or QA; team lead or training experience preferred. 

- Strong judgment on edge cases, with the ability to explain decisions clearly. 

- Clear communication and the ability to give constructive feedback. 

 

Nice to have 

- Experience building evaluation sets, analyzing accuracy, or managing annotation quality. 

- Understanding of AI-assisted annotation workflows and associated risk controls. 

 

VPN and Location Verification: What You Need to Know 

 1. Application and onboarding: no VPN. Please do not use a VPN when you apply or go through onboarding.  

 2. Production: VPN required. Once production starts, you will need to use a VPN. Opera browser is recommended. (Opera is a free built-in VPN and needs no extra installation.) We will send you setup instructions before production begins. 

 3. Which VPNs are allowed. Standard consumer VPNs are fine. High-risk proxies are not permitted. 

 4. Periodic checks in Production. For administrative reasons, we may occasionally ask you to access AI Trainer platform without your VPN so we can check your unmasked IP. 

Location: United States (remote).

Similar jobs

View all jobs