MLOps pipelines clarification using several tools

Question

I study this image below as a MLOps student. It is not official, I did not take a course or anything likewise. This makes the effort self-indulgent. I do not look for the answer to my questions, but maybe some material for me to read.

MLOps

The prior pipeline is:

1. Train a model offline and either use package cloudpickle on AWS S3 for complex models or reproduce it manually with build up business rules;
2. Write a minimal context-event function which will be called on dockerized application by command CMD;
3. Upload locally built image to ECR and redirect traffic to respective API Gateway route;

So far, so good. This allows me use this notebook and deploy the docker image as intended. It correponds to Week 7 and 8 of this class. My questions are:

1. How can I create a Github Action based on the content of a repository with respective Dockerfile and *.py file?
2. How to setup both ElastiSearch and Kibana to collect the logs I create with library logging?
3. What is the connection of weeks 0 to 5 and weks 6 to 9?

Feel free to offend me based on SO policies of question writing.