Open Nav

Accelerating Hybrid Cloud GPU Environment For Data Science

NGC containers


Consumer Technology




Deep Learning Computer Vision


Lightricks is a research-driven company, which is why we’ve built a team with some of the top computer vision researchers in the country. ensures our highly qualified researchers are focused on building the industry-leading AI technology that we are now world renown for, instead of spending time on engineering, configuration and DevOps.
Ofir Bibi
Ofir Bibi
Head of Research at Lightricks

About Lightricks

Lightricks was founded in January 2013 by five entrepreneurs with comprehensive knowledge in the fields of computer graphics, image processing, machine learning and artificial intelligence with the vision to change the visual editing industry. Lightricks provides solutions for the mobile market, developing fun and powerful visual editing apps for content creators. The company’s award-winning apps are used by millions of smartphone users globally.


Lightricks is the creator of popular, award-winning photo and video editing apps. Since Lightricks launched its flagship product Facetune, the company has built an arsenal of powerful ML-driven content creation apps used by millions worldwide. With a dedicated team of 30 computer vision researchers, Lightricks has delivered the most advanced and innovative computer vision features that render in real time on the device, to speed up the editing process. The team has built camera motion effects, applied AI technology to build a depth map that turns the images into engaging posters, and a video engine capable of producing a range of composition effects.


Cutting edge computer vision innovation with hybrid-cloud GPUs resource Optimization

Lightricks’ advanced editing solutions are built on advanced computer graphics and computer vision knowledge. The Lightricks application suite experienced massive user growth, reaching millions of users globally.
The dedicated AI team had to grow its team by 3X to maintain its high standard of cutting-edge computer vision technology while delivering solutions to a rapidly growing user base. The ML infrastructure became impossible to manage. The valuable time of researchers was wasted on DevOps and MLOps overhead, configuring, managing and monitoring the growing consumption of GPUs. As the team expanded, data volume grew exponentially, and more compute resources were added, including DGX-1s and, multi-cloud GPUs. With dozens of new ML models launched every week, versioning and reproducibility became a nightmare. Lightricks sought a solution for versioning, managing models and automating runtime as well as a complete GPU and data automation infrastructure that could keep up with their growth.


Deploying to accelerate MLOps and unify model and data management

The dedicated AI team smoothly grew from 5 to 30 researchers with the help of’s unified data science management platform. The team has minimized the time spent on DevOps, MLOps, data management and versioning with automation features.

Now, the research team can focus on innovation, building cutting-edge machine and deep learning models. Lightricks leverages to manage, automate and track hundreds of machine learning models from research to production. With’s meta-scheduling technology the team is successfully running hundreds of monthly training pipelines and is utilizing all compute resources. With one click, distributes computation jobs across their DGX-1’s with automated cloud-bursting once on-prem resources have reached full capacity. In addition, all models and experiments are automatically documented, tracked and visualized resulting in peak team collaboration and productivity.

  • Hyper experimentation – launching 100s of experiments with tracking and model comparing
  • Monitored metrics – visualizing model accuracy, loss value, health metrics, and performance
  • Optimized resource management – Meta-scheduling capabilities automatically leverage DGX-1s, K8 clusters, and cloud  infrastructure
  • Centralized Data Hub – managing all data in a single place with data version control, and auto-caching of frequently used datasets
    for fast model training
  • One click to any AI framework – Launching any ML or DL framework on any compute resource with a native integration to NVIDIA
    GPU Cloud container catalog


Since using, the AI team seamlessly scaled their research team by 3X in a few months. has reduced the onboarding overhead, and has unified team collaboration for accelerated growth. The research team has reduced time spent on DevOps by over 80% with automated MLOps, and has accelerated model training with cnvrg’s automated data caching and NFS optimization solution.

As a result, Lightricks ML innovation has scaled, and the team has deployed 3X more computer vision models to its applications, and have kept up with the millions of user requests. Lightricks has saved hundreds of thousands USD in GPU infrastructure costs and has saved up to 80% in cloud costs with spot instances.

Lightricks dedicated team of AI researchers has raised the bar on computer innovation, and has consistently ranked amongst the best of both Apple and Google App stores from 2013-2019.