Lightricks: ML recommendation models at scale

Qwak was brought aboard to enhance Lightricks existing machine learning operation, which was originally concentrated around image analysis, and enable fast delivery of complex tabular models.

Lightricks is a pioneer in innovative technology that leads to breakthrough moments throughout the creation process. On a mission to push the limits of technology to reimagine the way creators express themselves, the company brings a unique blend of cutting-edge academic research and design to every user experience.Qwak was brought aboard to enhance Lightricks's existing machine learning operation, which was originally concentrated around image analysis, and enable fast delivery of complex tabular models. The main requirement was to reduce the engineering dependency to a minimum along with providing full model training and deployment flexibility.‍

About Lightricks

Lightricks develops video and image editing mobile apps, known particularly for its selfie-editing app, Facetune.

Industry

Consumer Apps, Photo Editing, Content and Social Networks

Use Case

Recommendation Engine

Model Frameworks

Pytorch

From the get go, it was clear that Qwak understand our needs and requirements. The simplicity of the implementation was impressive.

Shaked Zychlinksi
Head of Recommendations Research
CT/CD

Continuous Training & Continuous Deployment

0

Engineering Dependency

Full

Model Analytics & Monitoring

Challenges

Developing recommendations engine models required intensive engineering support and building new infrastructure layers

  • Lightricks ML traditionally revolves around image analysis. The team needed to build ML models based on tabular and dynamic data which requires daily training on fresh data.
  • The platform's flexibility was crucial. Lightricks did not want to be locked down to a specific model structure and also needed a centralized way to train, deploy and access models at scale.
  • Lightricks knew that building Infrastructure requires a lot of effort, experience and time. (Build vs Buy blog)
  • Other vendors could not address the requirements of ease of use, speed of onboarding, continuous training & deployment and feature store support.

Solutions

Developing recommendations engine models required intensive engineering support and building new infrastructure layers

  • Lightricks ML traditionally revolves around image analysis. The team needed to build ML models based on tabular and dynamic data which requires daily training on fresh data.
  • The platform's flexibility was crucial. Lightricks did not want to be locked down to a specific model structure and also needed a centralized way to train, deploy and access models at scale.
  • Lightricks knew that building Infrastructure requires a lot of effort, experience and time. (Build vs Buy blog)
  • Other vendors could not address the requirements of ease of use, speed of onboarding, continuous training & deployment and feature store support.