Andrii B.

Developed an automated image recognition and collection platform for a US-based media client. The system parses images from social media sources, processes them through an ML pipeline, and automatically labels each image with visual attributes (e.g., hair type, style characteristics). The solution eliminated manual image selection, unified data storage, and enabled scalable image classification without duplicates. A similarity search (“look-alike”) feature was implemented using vector embeddings, allowing users to generate new selections based on reference images. The platform achieved high model accuracy and significantly increased processing throughput.