Join Daniel (ex-YouTube, now a founder at Superlinked) to figure out how to use vector embeddings to build engaging, explainable and safe software – both if you are a generalist engineer looking to get started or seasoned veteran looking for more advanced tips and tricks.
For years, data engineers have understood that it is a good idea to de-duplicate and reconcile your user and customer data and put them all in a central place – this way, you reduce errors, avoid work duplication and create a more consistent user experience. The billion dollar Customer Data Platform industry brought these ideas to scale.
The same transformation is currently underway in the Machine Learning stack – companies are centralising the models of their users for pretty much the same benefits. But it’s happening through trial and error, every company figuring this out on their own.
In this talk, we will look at:
1. Vector embeddings – why use them to understand your users
2. Vector Ops – the art of bringing embedding-powered systems to production
3. How user modeling serves as a foundation for recommender systems, bot detection and everything in between
4. Lessons learned from deploying the Superlinked framework in social platforms and marketplaces
Technical level: Introductory & Technical practitioner.