Case Study: Lightricks achieves 3× research-team scaling and 80% DevOps time savings with Intel Tiber AI Studio

Accelerating Hybrid Cloud GPU Environment For Data Science

Lightricks, the maker of award‑winning mobile photo and video editing apps, was struggling to scale its computer‑vision research as the AI team expanded from 5 to 30 people. Managing hybrid‑cloud GPU resources (DGX‑1s and multi‑cloud), exploding data volumes, experiment/versioning overhead, and heavy DevOps/MLOps tasks was consuming researchers’ time. Intel Tiber AI Studio was brought in to provide a unified data‑science management platform and automated MLOps to streamline GPU and data orchestration.

Intel Tiber AI Studio implemented automated meta‑scheduling, one‑click distribution and cloud‑bursting, centralized data/version control, experiment tracking, and automated data caching so Lightricks could run hundreds of monthly training pipelines and launch hundreds of experiments. The result: Lightricks scaled its research team 3X, reduced time spent on DevOps by over 80%, deployed 3X more computer‑vision models, and saved hundreds of thousands of dollars in GPU infrastructure costs and up to 80% on cloud spend.

Lightricks

Ofir Bibi

Head of Research