BS Data Annotation Engineer (Computer Vision)

ABOUT COMPANY

SoftServe Business Systems was founded in 2003. We aim high; our mission is to lead the digital revolution in the FMCG industry. It means delivering the best products & services possible that empower businesses to grow and improve efficiency.

We are a product company, and this affects our day-to-day activities. Our team is highly involved in the client's needs, we value business expertise, and we take every step with extra carefulness. Among our clients are businesses like AB InBev, Unilever, JDE, PepsiCo, Henkel, and others.

SoftServe Business System's ideal candidate can share the company's values and become a reliable partner for the team.

Project and Role

You'll join the team behind our Image Recognition (IR) platform — a Computer Vision product used by FMCG brands and retailers to monitor shelf execution, planogram compliance, product availability, and pricing.

We're looking for a Junior/Middle Data Engineer to build and maintain data pipelines around the annotation process and ensure the quality of labeled datasets used to train ML models.

This is a hands-on individual contributor role focused on automation, data quality, and improving the annotation workflow.

Requirements

• Working knowledge of Python and experience writing and maintaining scripts for data processing, validation, and automation
• Basic experience with Computer Vision data, including bounding boxes, polygons, segmentation, or similar annotation formats
• Basic understanding of ML model training and how data quality impacts model performance
• Familiarity with, or willingness to quickly learn, annotation tools such as CVAT
• Ability to identify systematic data quality issues, not only technical errors
• Previous experience with data annotation, labeling QA, or data quality processes
• Good problem-solving skills and a proactive approach to improving existing processes
• Strong attention to detail and a quality-first mindset

Nice-to-have skills

• Familiarity with MLOps tools, such as DVC, MLflow, or Label Studio ML backend
• Experience with object detection projects, particularly YOLO, Detectron2, or similar frameworks
• Exposure to retail or FMCG domains
• Understanding of planograms, SKUs, shelf monitoring, or category management

Responsibilities

• Building and maintaining scripts and tooling to automate repetitive parts of the annotation process, including pre-labeling, model-assisted annotation, format conversion, dataset validation, and progress tracking
• Integrating annotation tools such as CVAT into the broader data pipeline
• Supporting human-in-the-loop workflows where ML models generate preliminary labels and annotators verify or correct them
• Identifying manual and inefficient parts of the annotation pipeline and proposing ways to automate or simplify them
• Establishing and maintaining quality control processes for labeled data, including double reviews, golden sets, spot checks, and inter-annotator agreement metrics
• Tracking and reporting data quality metrics and identifying issues before datasets reach model training
• Maintaining and refining annotation guidelines based on recurring errors and edge cases
• Escalating systematic labeling issues to PM and ML team, providing concrete examples and suggested solutions
• Collaborating with ML Engineers to understand model data requirements and translate model failure analysis into actionable data and annotation improvements
• Working with the annotation team lead on guideline updates and tooling improvements
• Reporting pipeline health and data quality issues to relevant stakeholders when they may affect recognition quality or client requirements