About the role

We are looking for a Senior Annotator who understands not just how to annotate images, but why. You need to grasp the full pipeline - from raw field image capture in the Netherlands, through NAS storage and cloud systems, into Label Studio, and out as training data that directly shapes model performance.

You will own annotation quality for assigned projects. That means reviewing work, enforcing standards, and being the technical link between the annotation team and the ML workflow. If something in the data is causing model issues, you should be the first to notice. This role reports directly to the director.

What you will do

• Own annotation quality end-to-end - review work, set standards, enforce them
• Mentor and train annotators, including running benchmark assessments before they go live on projects
• Reason about the data - understand how labelling decisions translate into model training outcomes
• Work across the full pipeline: NAS → Label Studio → cloud → model feedback loops
• Monitor daily throughput and quality metrics on Grafana; flag deviations early
• Coordinate with the director on timelines, delivery formats, and project requirements
• Write and maintain annotation SOPs for each crop type and task

What we are looking for

• 2+ years in data annotation, computer vision data ops, or an ML data role
• Genuine understanding of how label quality affects model training - not just theoretical
• Hands-on polygon or instance annotation experience (Label Studio preferred)
• Comfortable reading dashboard metrics and spotting anomalies
• Catches errors others miss - quality-first before speed
• Clear communicator who can justify annotation decisions and quality calls to the team

This role is closed

We are not recruiting for this position at the moment, so the application form has been taken down rather than left to collect notes nobody is reading. The description stays up so you can see the kind of work it involves.

About the role

This role is not open at the moment. When we do recruit for it, it is not a fresher role — it needs at least 6 months of hands-on data annotation experience (image/polygon/bounding-box labelling or similar) to apply. We will train you on our specific tools, crops, and standards, but you should already know what good annotation looks like. Experienced in medical-imaging annotation (radiology, pathology or cell imaging)? We especially want to hear from you as we expand into healthcare.

What matters is precision, patience, and proven annotation discipline. The polygons you draw around plants in Dutch field images are the exact ground truth that autonomous farming robots use to learn to see. If your annotations are accurate, the machine gets it right.

Compensation: ₹25,000/month, plus a performance bonus of up to ₹5,000/month (accuracy + throughput) after probation.

What you will do

• Annotate agricultural field images using polygon tools in Label Studio - marking individual plants, weeds, or diseased specimens per project specifications
• Meet daily throughput targets (240-320 images/day) while maintaining quality benchmarks
• Pass internal annotation assessments before going live on any project
• Participate in regular quality reviews and act on feedback from the Senior Annotator
• Follow structured SOPs precisely - consistency matters as much as speed
• Flag ambiguous cases rather than guessing - accuracy over volume

What we are looking for

• Minimum 6 months hands-on data annotation experience — no freshers for this position
• Familiarity with annotation tools (Label Studio, CVAT, or similar) and polygon/bounding-box labelling
• High attention to detail: you notice things others skip past
• Able to focus and work methodically for extended screen sessions
• Reliable and consistent - annotation is a precision discipline
• Open to feedback and willing to follow structured processes
• Comfortable working full-time on a computer in our Vikhroli office

This role is closed

We are not recruiting for this position at the moment, so the application form has been taken down rather than left to collect notes nobody is reading. The description stays up so you can see the kind of work it involves.