Data Annotation Operations Manager

About the Role

We are looking for Data Annotation Operations Managers to run VinFast’s annotation production floor. From late 2026 the operation grows to several hundred annotators working on site, producing the ground-truth data that our ADAS and autonomous driving programmes are trained and validated on - 3D cuboids on LiDAR point cloud, 2D boxes and segmentation, lane and landmark, occupant monitoring, and anonymisation.

This is not a project management role. You will spend your day on the floor, not in a planning tool. We require candidates who have personally annotated to a production quality standard and can still do it today.

The workforce is organized in three tiers: you, a layer of team leads, and the annotators. Building and sustaining that team lead layer is what makes a two-person operations team viable at this scale, and it is a core part of the job.

Where This Role Sits

Our Data Quality & Evaluation team owns the label taxonomy, class definitions and the final acceptance criteria. You own the operational guidance - the working rules that tell an annotator how to hit those definitions in practice - and the day-to-day adjudication of edge cases within them. Where the taxonomy itself is unclear, inconsistent or wrong, you escalate it, and you are expected to.

Key Responsibilities

Production Operations

• Plan and allocate work across the workforce; track throughput against the delivery schedule and escalate shortfalls early
  

• Set and maintain productivity norms per data type, based on what the work actually takes
  

• Manage attendance, shift scheduling and capacity; coordinate with Human Resources and the partner university on contracts and output-based compensation
  

Annotation Quality

• Run regular audits of annotated output and issue corrective actions - you inspect labels yourself, not only dashboards
  

• Measure and act on quality indicators: IoU against reference, inter-annotator agreement, gold-set performance, rework rate
  

• Adjudicate edge cases on the floor, and turn each ruling into written operational guidance rather than a one-off answer
  

• Identify recurring error patterns by annotator, by data type and by shift, and address the cause rather than the instance
  

Training & Team Development

• Onboard and certify new annotators against a practical standard before they touch production data
  

• Maintain training materials, practical tests and assessment criteria; re-run training whenever guidance changes
  

• Identify, train and support the team leads who supervise day-to-day work, and build a progression path for strong annotators
  

• Enforce on-site data security procedures without exception, and report weekly on throughput, quality, attendance and risk
  

• 4+ years in data annotation, of which at least 2 managing an annotation team or a production floor
  

• You have personally annotated to a production quality standard. Shortlisted candidates complete a short practical labeling exercise as part of the process.
  

• Hands-on with an annotation platform - CVAT, Label Studio, Scale or equivalent - including task setup, job assignment and progress tracking
  

• Working command of annotation quality measurement: IoU, inter-annotator agreement, gold-set audit, rework rate - you can read these numbers and say what to change
  

• Demonstrated experience managing 50 or more people, directly or through a team-lead layer
  

• Able to write clear operational guidance that annotators follow without supervision, and to keep it versioned as rules change
  

• Strong planning, tracking and reporting discipline; working analytical use of Excel or Google Sheets
  

• Clear written and verbal Vietnamese; sufficient English to work with technical guidelines and class definitions written in English
  

• Willingness to work on site during production periods
  

Nice to Have

• 3D annotation on LiDAR point cloud (cuboids, orientation, tracking across frames) - the most relevant single background for this role
  

• Experience in an annotation vendor, BPO or AI data services delivery environment
  

• Experience building a QC or team-lead layer from within an annotator population
  

• Experience running an output-based compensation scheme
  

• Experience operating in a secure or access-controlled annotation facility
  

• Familiarity with autonomous driving data and what perception models need from labels