Case Study: Lightricks builds scalable recommendation models with Qwak

Qwak used by Lightricks to introduce ML recommendation models

Lightricks, the company behind the Facetune mobile app and other video and image editing products, needed to expand its machine learning operations beyond image analysis to build recommendation models using tabular and dynamic data. The team faced challenges around daily retraining on fresh data, limited engineering resources, and the need for a flexible, centralized MLOps platform with feature store support, so they brought in Qwak.

Qwak implemented an end-to-end ML platform covering model training, real-time and batch serving, feature store access, and model monitoring. This enabled Lightricks to train from fresh data, deploy with canary rollouts, and monitor production models with less engineering dependency, helping the company deliver a complex recommendation solution in a remarkably short timeframe and scale modern AI/ML pipelines more efficiently.

Lightricks

Shaked Zychlinski

Head of Recommendations