Case Study: Ancestry accelerates historical record decoding with Labelbox

Ancestry speeds weekly model iteration with Labelbox

Ancestry, the genealogy company, faced a bottleneck in its machine learning pipeline. Their data science team needed to extract data from historical records like census documents faster, but producing high-quality training data and incorporating feedback from domain experts was slow and inefficient. This hindered their shift to a data-centric approach and optimization of their MLOps workflow.

To solve this, Ancestry adopted Labelbox's platform, utilizing its model-assisted labeling and collaboration features. This solution combined automated workflows with direct expert feedback within native editors, accelerating signal generation. Using Labelbox as a data engine, the team achieved a weekly model iteration cycle, saved significant time in collaboration, and was able to train and test new models in record time while maintaining high data quality.

Ancestry

Stanley Fujimoto

Senior Data Scientist