How to build end-to-end recommender systems at reasonable scale

Jacopo Tagliabue, RecSys expert and former Director of AI at Coveo, and Ronay Ak, Senior Data Scientist at NVIDIA Merlin team, join Hugo Bowne-Anderson, Outerbounds’ Head of DevRel, in a live coding session to show how to build a production-ready pipeline for deep learning recommendations.

In particular, you’ll learn how, as a single data scientist, ML engineer, or a small team of MLEs, you can train a cutting-edge deep learning model (actually, several versions of it in parallel), test it, and deploy it without any explicit infrastructure work, without talking to any DevOps person, without using anything that is not Python or SQL.

You’ll learn all this through a popular RecSys challenge, user-item recommendations for the fashion industry: given a shopper's past purchases, can we train a model to predict what he/she will buy next?

Our goal is to build a pipeline with all the necessary real-world ingredients:

• dataOps with Snowflake and dbt;
• training Merlin models (possibly on GPUs), in parallel, leveraging Metaflow;
• experiment and parameter tracking;
• serving cached prediction through FaaS and SaaS (AWS Lambda, DynamoDb, the serverless framework);
• error analysis and debugging with a Streamlit app.