Case Study
Faster recommendations on streaming app at low cost on Redshift for M&E conglomerate


An American M&E conglomerate was looking for a performance upgrade for its streaming app. Due to our proven expertise in migrating varied workloads to AWS, the client approached Agilisium to migrate their streaming platform’s R based recommendation engine to AWS cloud.


The recommendation engine suggests customers the shows they can next watch based on their viewing pattern. The existing engine was written in the R language. Agilisium’s AI& ML experts entirely refactored the engine for Apache Spark incorporating alternating Least Squares (ALS) recommender & Jaccard Similarity index to arrive at the recommendations requested by the client.

Solution Highlights

The client chose to leverage the low storage costs of AWS and move their viewership data from various partners such as Hulu, YouTube, Apple TV to AWS Redshift. This viewership data was processed by the Spark-based recommendation engine to arrive at the top 10 recommended shows. These recommendations are stored in DynamoDB – the fully managed serverless DB service that can automatically scale to service the unpredictable, enormous volume and velocity of queries that are natural to a streaming app. This enables the delivery of lightning-fast recommendations and personalized offers to app users.


Results and Benefits
  • Users get faster recommendations each time they open the app
  • Lesser storage costs