MLOps Concepts Advanced

Warning

Below is a poorly structured text that is not readable yet.

Batch Inference Architecture

TODO classification of goods by name in cash receipts

We will return to batch inference systems later when we look at ETL processes.

Embedded Model Architecture

TODO Face detection in an image while taking a photo

Real-Time Streaming Architecture

TODO recognition of car numbers in the stream

OLAP vs OLTP

ML product boundary

Data Extraction vs Data Integration

ML dev process. ML Pipeline and ML infrastructure

More is needed, including:

  1. СI/CD

  2. Data Handling

  3. ML pipelines

  4. More Computing Power