ML/OpenCV Data Labeler

About the Role

We’re looking for a Machine Learning / Computer Vision Data Labeler to support customers onboarding and build high-quality training datasets for our computer vision products used in manufacturing environments. This role sits at the intersection of ML data operations and light product/customer work—you’ll help us understand what customers do on the factory floor, collect and analyze representative sample data from each station, and translate real-world processes into clear labeling instructions and reliable datasets. This is not a super-senior role, but it does require strong ownership, attention to detail, and comfort working with highly confidential customer data.

Responsibilities

• Coordinate and execute sample data capture across all manufacturing stations, ensuring coverage of real-world variation
• Work with our on-site implementation team to validate camera setup outputs (camera position, field of view, recording settings, connectivity, sample clips/images).
• Organize, clean, and curate datasets (images/video), including selecting representative samples, filtering unusable footage, and documenting capture conditions.
• Perform data labeling/annotation for computer vision tasks (e.g., classification, object detection, segmentation, defect tagging, action/process step labeling—depending on the use case).
• Create and maintain labeling taxonomies and annotation guidelines that are consistent, scalable, and easy for others to follow.
• Run quality checks (spot checks, consistency reviews, edge-case handling) and partner with ML/Engineering to continuously improve label quality.
• Conduct lightweight exploratory analysis on incoming datasets (e.g., distributions, coverage gaps, common failure modes, ambiguity hot-spots).
• Flag data issues early (missing stations, misaligned camera views, insufficient examples, inconsistent definitions) and propose fixes.
• Provide structured feedback to ML and product teams: what data we have, what we’re missing, and what will improve model performance.
• Support customer onboarding by learning what the client does, mapping their workflow/stations, and translating their needs into data/labeling requirements.
• Communicate clearly with internal stakeholders and occasionally with customers to align on labeling definitions, success criteria, timelines, and data handling expectations.
• Document processes, station definitions, and dataset decisions so teams can move fast and stay aligned.
• Work with sensitive/secret customer manufacturing data and follow strict security policies (access control, secure transfer/storage, need-to-know practices, and customer-specific handling requirements).

Qualifications

• 1–4 years of experience in a role involving data labeling/annotation, ML data operations, computer vision datasets
• Working knowledge of computer vision fundamentals (classification vs detection vs segmentation; what labels are used for; why consistency matters).
• Experience with labeling tools such as CVAT, Labelbox, V7, Supervisely, or similar (or the ability to learn quickly).
• Comfort working with data formats/workflows (e.g., CSV/JSON annotations, COCO-style formats, dataset folders, basic versioning concepts).
• Strong written and verbal communication skills; able to explain labeling decisions and customer workflows clearly.
• Professional maturity and discretion—ability to handle highly confidential customer data.
• German language ok, strong communication in English preferred

Nice to Have

• Exposure to manufacturing environments (industrial processes, station-based workflows, quality inspection).
• Familiarity with camera systems / video capture pipelines (e.g., frame rate, resolution trade-offs, lighting impacts, field of view).

Offer Details

• Full-time employment via temporary agency (EoR)
• Remote only, full-time dedication (40 hours/week)
• EU Timezones
• Competitive compensation package.
• Opportunities for professional growth and career development.
• Dynamic and inclusive work environment focused on innovation and teamwork