Case
Media Recommendations Engine for NBCUniversal’s Hayu
Who are Hayu?
NBCUniversal’s Hayu service is a fast-growing video streaming service. They wanted to personalise content recommendations to increase content consumption. To achieve this, Merapar delivered a personalisation service based on advanced Machine Learning Algorithms.
Challenge
The customer required a CMS to configure the content including search and recommendation settings that would personalise user content to increase content consumption, improve customer satisfaction with the service whilst also providing additional Business Intelligence including consumer habits, click-through and conversion rates.
Solution
Merapar created a serverless CMS built on a number of AWS services including AWS Amplify, AWS Lambda and Amazon Kinesis Data Firehose. AWS Personalize was then provisioned to provide curated content and intuitive suggestions. The solution also connects to the wider Content Management System (CMS) in order to enable content placement across the user catalogue.
Results
The initial POC was completed in three weeks with first deployment in approximately twenty weeks. Merapar has subsequently delivered a comprehensive content discovery solution for the Hayu streaming service to NBCUniversal, now operating in 40+ countries. This solution serves both end customers and editorial use cases. Personalisation is a high priority and our solution uses customer data to drive “continue watching” functionality, personal playlists and targeted recommendations.
Related
CaseDevice Migration for a Cable Company
Merapar built a serverless migration on AWS that moved over 5 million of a cable company’s devices, with new firmware and customer data, to a new platform.
Read the case
WhitepaperHow to harness the power of AWS Personalize
Most OTT services want recommendations to lift viewership and catalogue breadth. How to deliver them quickly, at scale and with ease.
Read the whitepaper
BlogPure serverless machine learning inference with AWS Lambda and Layers
Running scikit-learn inference purely on AWS Lambda, with the model in its own layer, and how the cost compares with a permanently hosted SageMaker endpoint.
Read the post
Could we do the same for you?
Tell us what you are trying to build, and we will tell you how we would approach it.