As an Image Annotator at Verisure, you will create and validate high-quality training and evaluation datasets for our on-device image detection models. Your work directly contributes to improving model accuracy, robustness, and user trust through precise labeling and quality assurance.
What You'll Do
- Annotate images with precision (e.g., bounding boxes and classes/attributes) using labeling tools and taxonomies
- Apply clear guidelines and edge-case rules; raise ambiguities and propose refinements to keep label definitions crisp
- Quality-assure labels through spot checks, inter-annotator agreement checks, and consensus workflows; fix issues at speed without compromising quality
- Curate datasets for model training/validation/test, corner cases, and rare classes to improve coverage
- Support model evaluation by producing ground-truth labels for benchmarks and release testing; help compute and sanity-check key metrics
- Protect data privacy & security by following data-handling procedures at all times
What We're Looking For
- Strong visual attention to detail and consistency
- Ability to follow written guidelines and give clear, constructive feedback when something is ambiguous
- Comfortable with repetitive tasks when needed—while keeping quality high
- Good written and spoken English
- Availability 10–20 hours per week during the semester; schedule can flex around exams
Nice to Have
- Swedish is a plus, not a requirement
Technical Stack
- labeling tools
- taxonomies
Benefits & Compensation
- Impact you can see: your labels and insights will measurably improve production model performance
- A supportive team that values clarity, inclusion, and continuous learning
- Flexible hours adapted to your studies and exam periods
- Competitive hourly compensation
Work Mode
Flexible hours adapted to your studies and exam periods
