Under Guidence of

░v░i░m░a░l░ ░d░a░g░a░

Now a quick review about project :

This is a project which would automate everything after developer commits code in GIT
Prerequities: GIT , Docker and Jenkins should be pre-installed on the top of Base system RHEL8

To deploy the project

• Create a file named "Dockerfile" and copy paste from Dockerfile then save.
• Run the command same as docker build -t mlops:latest .
• In GIT bash go to Your project_folder/.git/hooks//
• Create post-commit file and copy paste data from post-commit
• Now in jenkins create job1 and write cp -v -r -f * /myweb in Execute shell.
  set triggers for SCM pooling and save
• Create job2 and copy paste the code
  
  if sudo python3 /myenv/checking.py == CNN model
then
if sudo docker ps -a | grep testing
then
sudo docker rm -f testing
sudo docker run -v /myweb:/MLops --name testing mlops:latest
else
echo "Image not present"
fi
fi
  set triggers to build after job1
• Create job3 and copy paste the code
  
  sudo python3 /myenv/mail.py
  set triggers to builtd after job2
• Setup job4 as pipeline for all 3 jobs using Build pipeline as a monitoring system.

Refer ss for further clarifications

• Job1 : Download the code and store it in /myenv folder.
  
• Job2 : Check the code and accordingly launch the container for model training. Also it check the accuracy of the model and if acc < 0.95 it retrains the model after changing the architecture of the model. After the training the model it stores the accuracy of the model to /MLops/accuracy.txt file.
  
• Job3 : After getting best accuracy it notify to the developer for accuracy via E-mail.
  
• Job4 : Monitoring system for all 3 jobs using Build pipeline.
  

I would like to Thank Mr. Vimal Daga sir for this Intresting task.