About the role

NorthLadder runs device trade-in and recommerce across the Middle East, Europe and other regions. We’re building a computer-vision model that grades the cosmetic condition of used phones, tablets and laptops from photos and video. You’ll produce the labelled data that trains and tests that model. That means marking every scratch, crack and dent precisely, following a detailed grading guide. This is a 3-month project to build and validate the model’s first full training dataset.

What you’ll do

- Label device images and video frames with bounding boxes, polygons and segmentation masks for defects such as scratches, scuffs, cracks, dents, chips, screen burn-in, dead pixels and discolouration
- Give each device zone (screen, back, frame, corners, camera) a cosmetic grade using NorthLadder’s grading rubric
- Review and correct labels that were pre-generated by our models, and note where the model is consistently wrong
- Flag images that are blurry, have glare, are badly framed or are ambiguous, and escalate edge cases to the QA lead




- Take part in consensus and QA rounds, such as double-labelling and resolving disagreements between labellers
- Meet agreed accuracy and throughput targets (e.g. ≥98% QA pass rate)

What we’re looking for

- Able to start within 2 weeks and be productive within the first week of onboarding
- 1–3 years of image annotation experience, ideally on computer-vision projects (object detection or segmentation)
- Hands-on experience with tools such as CVAT, Label Studio, Roboflow, Labelbox or similar
- Strong attention to detail, with the ability to spot small differences in surface texture, colour and reflection
- Able to follow detailed written guidelines consistently over long, repetitive sessions
- Valuable written English, enough to document edge cases clearly
- Comfortable working with quality metrics and feedback loops

Interested candidates can share their profile at [email protected]

📌 Data Labelers (India)
🏢 NorthLadder
📍 India