DEV Community

Discussion on: Learnings from creating recommendation engines with Amazon Personalize

Collapse
 
belkheir profile image
Mahamed Belkheir

This is really valuable, especially for small teams and people without experience in the field having to Implement recommendation systems, I've always been wary of systems that claim "no experience needed", sure you can use it but how much value you're getting depends on how far you are from 0

Collapse
 
cremich profile image
Christian Bonzelet

I am totally with you.
Machine-Learning based recommender system is a very complex topic. And if you want to fine-tune your recommendations it is good to have some foundational skills in your bag.

Therefore we also decided to centralize the knowledge of recommendation engines in a team that will support our other teams in the future.

I recently had a chat with a data scientist from AWS. And he also mentioned, that Personalize is often a good long-term choice. From his experience, the quality from the pretrained model in Personalize is really impressive and often better compared to if you aim to build your own solution using Amazon SageMaker for example.