The Project Manager is responsible for ensuring the successful planning, execution, tracking, delivery and closure of client projects.
Responsibilities
Project planning, scoping, requirements gathering and validation with client
Identifying and working with the relevant client and internal stakeholders to plan and resource projects appropriately in terms of quality, cost, and schedule requirements
Risk analysis and contingency planning (plan-do-check-act cycle)
Leading the internal and external team to successful execution and delivery of client projects
Creating schedule and monitoring timeliness of delivery
Requirements
Bachelor’s degree (B.A.) from a college or university in related filed
Leads a program of AI training data projects at scale. Owns program-level strategy and delivery, manages a cross-functional team, and is the primary owner of the client relationship for the program.
Responsibilities
Define program goals, roadmap, milestones, and operating model across multiple projects and vendors
Lead a delivery team (PMs, Coordinators) with clear goals, capacity plans, coaching, and performance feedback
Act as the day-to-day executive contact; run governance, planning, and QBR-style reviews; align scope, priorities, and change
Co-own quality outcomes with the Manager of Quality; align on guidelines, audits, acceptance criteria, and corrective actions
Plan and secure rater/annotator capacity; optimize shifts, handoffs, automation, and tooling to meet high-volume targets
Support the day-to-day quality and accuracy of digital content across online platforms by completing standardized content curation and metadata validation tasks, following defined guidelines and checklists, and documenting issues for timely resolution under regular supervision.
Responsibilities
Perform daily content curation activities for written publications and audio content using established SOPs, checklists, and client guidelines
Validate content and metadata for completeness, consistency, and basic accuracy
Identify and document content issues, defects, and bugs in internal tools using required templates and ticketing standards
Escalate unclear, high-impact, or out-of-policy issues to the Team Lead/SME with clear context and supporting evidence
Manage assigned queues to meet defined throughput expectations and turnaround times/SLA requirements
Review annotators' ACCEPT/REJECT decisions and reject reason codes, with a focus on preventing false matches and false rejects.
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
Requirements
Nice to have: Building evaluation sets
Nice to have: Analyzing accuracy
Nice to have: Managing annotation quality
Nice to have: Understanding of AI-assisted annotation workflows
Review pairs of e-commerce listings and decide whether they are the same product and the same sellable variant, returning ACCEPT or REJECT.
Responsibilities
Review pairs of e-commerce listings and decide whether they are the same product and the same sellable variant, returning ACCEPT or REJECT.
Compare key attributes such as brand, model number, size/capacity, color, quantity, and set composition, using product names, images, and listing pages, including Korean text embedded in images.
For REJECT decisions, select the correct reason code(s) from the provided code list.
Handle special cases per project rules, such as missing or blocked pages, and variant options shown as swatches with none pre-selected.
Record edge cases, uncertainties, and observations in the notes field.
Review annotators' ACCEPT/REJECT decisions and reject reason codes, with a focus on preventing false matches and false rejects.
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.
Requirements
Nice to have: Experience building evaluation sets
Nice to have: Analyzing accuracy
Nice to have: Managing annotation quality
Nice to have: Understanding of AI-assisted annotation workflows
Review pairs of e-commerce listings and decide whether they are the same product and the same sellable variant, returning ACCEPT or REJECT.
Responsibilities
Review pairs of e-commerce listings and decide whether they are the same product and the same sellable variant, returning ACCEPT or REJECT.
Compare key attributes such as brand, model number, size/capacity, color, quantity, and set composition, using product names, images, and listing pages, including Korean text embedded in images.
For REJECT decisions, select the correct reason code(s) from the provided code list.
Handle special cases per project rules, such as missing or blocked pages, and variant options shown as swatches with none pre-selected.
Record edge cases, uncertainties, and observations in the notes field.
Review pairs of e-commerce listings and decide whether they are the same product and the same sellable variant, returning ACCEPT or REJECT.
Responsibilities
Review pairs of e-commerce listings and decide whether they are the same product and the same sellable variant, returning ACCEPT or REJECT.
Compare key attributes such as brand, model number, size/capacity, color, quantity, and set composition, using product names, images, and listing pages, including Korean text embedded in images.
For REJECT decisions, select the correct reason code(s) from the provided code list.
Handle special cases per project rules, such as missing or blocked pages, and variant options shown as swatches with none pre-selected.
Record edge cases, uncertainties, and observations in the notes field.