AI Data Pipeline & Annotation Tooling (Python + TypeScript)
Built and maintained data labeling pipelines and internal annotation tools using Python and TypeScript to support machine learning workflows. Developed backend services in Python (FastAPI) to process, validate, and store annotated datasets, and created frontend interfaces using TypeScript (React) for efficient labeling and review workflows. Handled multiple data types including text (NER, classification) and images (bounding boxes). Automated preprocessing, dataset validation, and format conversion (COCO, JSON, CSV), reducing manual effort by 40%. Implemented quality control systems such as validation scripts, consensus scoring, and review dashboards to maintain over 98% annotation accuracy. Worked closely with ML engineers to ensure datasets were optimized for training pipelines, and deployed scalable solutions using Docker-based environments.