
VinDynamics designs humanoid robots focused on home use, offering safe and affordable assistance for daily tasks. The company combines research-grade hardware with intelligent software to support…

VinDynamics designs humanoid robots focused on home use, offering safe and affordable assistance for daily tasks. The company combines research-grade hardware with intelligent software to support…
Your next opportunity is in here somewhere. Sign up to explore 52,000+ startups and their open roles. No spam. No gamification. Just jobs.
52,000+
Startups
65,000+
Open Roles
1,500+
New This Week
Role Overview
VinDynamics is developing general-purpose humanoid robots
that learn from large-scale data collected in simulation, teleoperation, and the real world. We are seeking a Head of Data Collection & Management
to own the end-to-end data lifecycle
that powers AI models for humanoid robots.
This role is responsible for how data is collected, structured, validated, stored, and reused
across the humanoid AI stack—ensuring data quality, scalability, traceability, and alignment with learning objectives (VLA, RL, diffusion, flow-matching). You will turn data into strategic assets
.
Scope of Ownership
The Data Collection & Management Division owns:
Key Responsibilities
1. Data Strategy & Architecture
Define the data strategy and roadmap
for humanoid learning across: Perception, Manipulation, Autonomy, Foundation models (VLA, multimodal policies)
Design scalable data architectures
covering ingestion, storage, indexing, and retrieval.
Define data standards: schemas, metadata, formats, timestamps, synchronization.
2. Data Collection Systems
3. Dataset Curation & Quality
Own dataset cleaning, filtering, balancing, and validation
4. Infrastructure, MLOps & Governance
Build and maintain data infrastructure
supporting large-scale training and evaluation.
Integrate with ML pipelines, experiment tracking, and model registries.
Ensure data governance
, access control, security, and compliance.
Define policies for data reuse, privacy, and long-term storage.
5. Team Leadership & Cross-Functional Execution
Build and lead a multidisciplinary team
Annotation / tooling engineers
Coordinate closely with: AI Manipulation, Autonomy, Simulation & Hardware teams
Translate model requirements into data collection plans and KPIs
.
Required Qualifications
Preferred Qualifications
Why Join VinDynamics
Own the data backbone
of a humanoid robotics platform.
Direct impact on how robots learn, adapt, and scale.
Work at the intersection of robotics, AI, and real-world systems.
Competitive compensation, leadership growth, and long-term impact.
Lead development of data collection pipelines
from:
Teleoperation systems (VR/XR, leader–follower, motion capture)
Simulation (Isaac Sim / MuJoCo / custom)
Real humanoid deployments from both robots and environment
Ensure consistent capture of:
Sensor data (RGB, depth, tactile, force, IMU)
Actions, states, rewards, and task context
Language instructions and annotations (for VLA)
Enable scalable, repeatable, and safe data collection operations.
.
Define data quality metrics: coverage, diversity, noise, bias, failure cases.
Build tooling for:
Auto-labeling and semi-supervised annotation
Dataset versioning and lineage tracking
Train / eval / benchmark splits
Work closely with AI teams to align datasets with model performance gaps
.
Bachelor’s / Master’s / PhD in Robotics, Computer Science, Data Engineering, or related fields.
Strong experience building data pipelines for robotics or embodied AI systems
.
Deep understanding of how data supports:
Reinforcement learning
Imitation learning
Multimodal / VLA models
Hands-on experience with large-scale data systems (on-prem or hybrid).
Proficiency in Python; familiarity with distributed systems and ML workflows.
Proven experience leading technical teams and cross-functional projects.
Experience with humanoid or mobile manipulation robots
.
Background in teleoperation, motion capture, or human demonstration systems
.
Familiarity with sim-to-real data strategies
.
Experience supporting foundation models
for robotics.
Experience transitioning data systems from R&D to production
.