Building an ETL for Black Box Training

One of the actual problems is training ML/DL systems on real data.

Based on the experience of work in a successful project, the speaker will tell:

— What are the peculiarities of working with physical documents (driver's license, passports, etc.) from the point of view of ML/DL-systems.
— Why we need DWH and ETL, even if we have a "neuron that does its own thing".
— What are the peculiarities of data preparation for ML/DL systems.
— Why it makes sense to read books about projects from the USSR of the 60s.