Deliver viewing experiences for super fans with Amazon Personalize
summary
Demo uses MovieLens 20M (28,000 movies, 138,000 users, 20M interactions) to show Amazon Personalize trained from CSVs in S3 and served via GetPersonalizedRanking with real‑time PutEvents feedback. Pulselive injects a Lambda proxy behind CloudFront + API Gateway, logs A/B test data to DynamoDB, and reports a sustained 20% uplift in video views over three weeks against control (goal was 10% significance). Features demonstrated include cold start handling, item exploration, and use of GetPersonalizedRanking for next‑watch recommendations.
AI-written summary · about
chapters (8)
- 0:00Introduction and overview
- 2:31Industry challenges in media
- 5:45Amazon Personalize explained
- 7:51Demo of Personalize features
- 11:49Pulselive’s use case
- 16:12Implementation steps
- 20:36A/B testing results
- 22:52Future plans and resources
related talks
26mPersonalisation at scale with AI/ML and Amazon Personalize
27mHow Warner Bros. Discovery uses Amazon Personalize
27mTransform online shopping experiences with personalized recommendations
03mMachine Learning for Personalization
01h 21mUnlocking Personalization: A Deep Dive into Modern Recommendation Algorithms
45mReal-time, Streaming Advanced Analytics....
24mAI-Powered Transformation in Fashion E-Commerce
19mX1 Recommendations Platform: Started from the bottom..