ML Engineer: Data Labeling & Automation

Dyson; Dubai; On-site; AED 180,000 - 280,000; Full time

Job summary

Dyson is seeking a Data Intelligence Machine Learning Engineer to design in-house tools that automate data labelling Pipelines, reducing manual annotation by using Active Learning, Weak Supervision, and synthetic data techniques. You will bridge raw data collection and model-ready datasets to enable scalable, high-quality labeling.

You will build end-to-end labelling architectures, HITL interfaces, and integrate with our data lakes and ML training infrastructure while collaborating with Data

Qualifications

• Experience designing data-centric AI or data labeling pipelines.
• Proficiency with Python ML stack and data processing tools.
• Experience with Active Learning or Weak Supervision strategies.
• Experience with large-scale unstructured data and cloud ML services.

Responsibilities

• Architect automated labelling pipelines using frameworks like Snorkel or custom loops.
• Build HITL interfaces where humans intervene on uncertain samples.
• Implement QA and denoising to improve data quality.
• Integrate labelling tools with data lakes and ML infrastructure.
• Fine-tune teacher models for high-quality pseudo-labels.

Skills

Python

PyTorch/TensorFlow

Active Learning

Weak Supervision

Data Engineering

SQL/NoSQL

Cloud Platforms (AWS/GCP/Azure)

DVC

MLOps

Data Visualization (Tableau/Power BI/J

Education

Bachelor's or Master's degree in Computer Science/Engineering/Data Science

Tools

Snorkel

Cleanlab

Tableau

Power BI

Job description

Dyson is seeking a Data Intelligence Machine Learning Engineer to design in-house tools that automate data labelling Pipelines, reducing manual annotation by using Active Learning, Weak Supervision, and synthetic data techniques. You will bridge raw data collection and model-ready datasets to enable scalable, high-quality labeling.

You will build end-to-end labelling architectures, HITL interfaces, and integrate with our data lakes and ML training infrastructure while collaborating with Data