Job description
We are looking for a Quality Analyst to support the ongoing improvement of our Optical Character Recognition (OCR) systems. In this role, you will work directly with our OCR Quality Check platform—a web-based interface designed to display model predictions side-by-side with source images. You will be responsible for reviewing, verifying, and correcting model outputs to help create high-quality datasets for model training and evaluation.
This position is ideal for someone detail-oriented, organized, and motivated to contribute to AI model quality through structured data review and annotation.
What You’ll Be Doing
Use the OCR Quality Check interface to review and validate OCR model predictions (e.g., digits, serial numbers, manufacturer names, and special characters).
Accept, correct, or reject model predictions based on their accuracy compared to the original image.
Identify and log issues such as mispredictions, misclassifications, and bounding box errors.
Ensure annotation consistency across a variety of meter types and environments.
Tag and document edge cases to support model improvement.
Contribute to the creation of high-quality labeled datasets used to retrain and refine our OCR models.
Generate error reports and accuracy summaries to assist the AI and product teams.
Collaborate with ML engineers and product stakeholders to provide insights on common model failure points.
Why This Role Matters
Directly improves model accuracy by identifying and correcting flawed predictions.
Helps the AI team ensure the OCR system generalizes well across real-world scenarios.
Enables continuous model improvement through human-in-the-loop feedback.
Ensures the quality of datasets used for future model training iterations.
Key Platform Features You’ll Use
Batch image upload and review tools.
Side-by-side comparison of images and OCR model predictions.
Annotation and correction tools for different fields (e.g., digits, serials, manufacturers).
Tools for error tagging, reporting, and dashboard analytics.
Centralized storage of all reviewed and verified data.
If you're passionate about contributing to the development of cutting-edge AI technologies through high-quality data curation, we'd love to hear from you.