Class_Notes.txt

MLOps: Edu 

Start Date: 29th Nov. 2025
8:30PM-11:30PM IST - SAT-SUN

####################
Day1: 29th Nov. 2025
####################

	- MLOps Introduction :::
	
		Github Account :
		
			https://github.com/
			
			
	- What is DevOps ???
	
	SDLC : Software Development Life Cycle :
	
		- Requirement Analysis 
	
		- Designg & Document 
		
		- Code & Development 
		
		- Testing 
		
		- Implement to Prod
		
		- Monitor/Maintain 
		
		
	- Waterfall: 
		- It is Linear in fashion.
		- It strictly follows Top-Down Approach
	
		- Requirement Analysis 
	
		- Designg & Document 
		
		- Code & Development 
		
		- Testing 
		
		- Implement to Prod
		
		- Monitor/Maintain 	
		
		
	Software Applications :
	
		- Desktop Applications
		- Web Applications
		- Mobile Applications
		- Embedded Applications
	
	
			
		- Desktop Applications	:::
			
			SuperMarket Billing System:
					--> Functions/Modules 
							- User Interface Design 
							- Inventory Mgmt 
							- Billing Mgmt
							- Supplier Mgmt 
							- Payment Mgmt 
								- Cash Payment 
								- Card Payment
							
										- Online Payment
										- UPI Payment 
							
					
	Core Project :
	
		- Requirement Analysis 		--> 12 Months 
	
		- Designg & Document 
		
		- Code & Development 		--> 6th Month 
		
		- Testing 
		
		- Implement to Prod
		
		- Monitor/Maintain


	Enhancement Project :
	
		- Requirement Analysis 		
	
		- Designg & Document 
		
		- Code & Development 		
		
		- Testing 
		
		- Implement to Prod
		
		- Monitor/Maintain
		
		
	AGILE Methodologies :::
	
		- Desktop Applications	:::
			
			SuperMarket Billing System:
					--> Functions/Modules --> iterations  
							- User Interface Design 
							- Inventory Mgmt 
							- Billing Mgmt
							- Supplier Mgmt 
							- Payment Mgmt 
								- Cash Payment 
								- Card Payment		

							
										- Online Payment
										- UPI Payment 
										
	Iteration 1 : User Interface Design 
	
		- Requirement Analysis 		
	
		- Designg & Document 
		
		- Code & Development 		
		
		- Testing 
		
		- Implement to Prod with proper manual approvals
		
		- Monitor/Maintain
		
		
	Iteration 2 : Inventory Mgmt
	
		- Requirement Analysis 		
	
		- Designg & Document 
		
		- Code & Development 		
		
		- Testing 
		
		- Implement to Prod with proper manual approvals
		
		- Monitor/Maintain
		

	nth Iteration : Online Payment
	
		- Requirement Analysis 		
	
		- Designg & Document 
		
		- Code & Development 		
		
		- Testing 
		
		- Implement to Prod with proper manual approvals
		
		- Monitor/Maintain
		
		
	Using AGILE Methodologies:
	
	 - We can Achieve :
	 
		- Continuous Development
		- Continuous Integration 
		- Continuous Testing 
		- Continuous Delivery
			- Used to perform Production release with manual approvals.
		
	 - We cannot Achieve :
	 
		- Continuous Deployment 	
			- Used to perform Production release without any manual approvals.
			
			
	- What is DevOps ???
	
	- DevOps :::::
	
		DevOps is a Software Development Strategy, which helps to promote the collaboration between the teams like Development Team and Operations Team to achieve Continuous Development, Continuous Integration, Continuous Testing, Continuous Delivery, Continuous Deployment and Continuous Monitoring in more Automated Fashion.
		
			Continuous Delivery & Deployment:
			
				- Both Continuous Delivery & Continuous Deployment are used to release the changes from Non-Prod to Prod.
				
				Continuous Delivery:
				
					- Continuous Delivery Exepects Manual Approvals
					- We might experience the Application Down-time
	
					Eg: 
				
					Banking Domain :
					
						Online Banking / Credit Card Transaction
				
							Create a Production Release Window:
								4 - 6Hrs 
								
	
				Continuous Deployment:
				
					- Never Exepects Manual Approvals	
					- Deploy the changes to production without any down-time.

					Eg:
					
					- Amazon/Netflix/Facebook/Google ::
					
						
				
	
					
	- How to Implement DevOps :::
	
		- Identify the Teams involved in the process of SDLC :
		
			- DevOps Team 
				- Infra-Structure Management Team 
				- Application Development Team
				- Testing Team 
				- Release Management Team 
				- Production Support Team 
				- Production Monitoring Team 
				- IT Security 
				
				
		- Environments :
		
			Non-Prod 									Prod Environments
			
				Dev Environment
				
				Test Environment
				
					QA 
					
					UAT 			================>		Production Servers - 1,2,3,4,5
	
		
		
		- Application Architecture :
		
			- Monolith Application Architecture:
			
				- It Tightly-Coupled Application Architecture
				- All the dependencies are tightly coupled 
				- With the we can achieve only Continuous Delivery.
				- We cannot achieve Continuous Deployment
			
			
			- Micro-Service Based Application Architecture: 
			
				- It is Loose-Coupled Application Architecture,
				- Each and every function is considered as Micro-Service.
				- With this we can achieve Continuous Deployment.
				
				
	Devops :::
	
		- People
		- Strategy/Process 
		- Tools
		
	
####################
Day2: 30th Nov. 2025
####################		
	
	MLOps Lifecycle :
	
	MLOps Usecase :
	
	Versioning :
		
		Version Control System using Git :
		
		Git Demo :
		
	ML Project and Packaging ML Model 
	
	CICD Pipeline:
	
		- SCM_Checkout 	==> VCS ?
		- Application Build
		- Create Application Artifacts 
		- Test 
		- Deploy 
		
		
	How to Implement MLOps :::

	
		- Data Engineers		-> Prepare the Source Datasets 
		
		- Data Scientists 		->  Explore the datasets, analyse, check the feasibility of datasets 
									Perform the training, tuning, and evaluate the Model
		
		- ML Engineers 			-> Deploy the ML Models to prod -> MLOps Strageies CICD, define best practises
									Should have MLOps Knowledge
		
		- Business Stakeholders -> Bring business values using this ML Models 
		
		- Data Security/Governance Team  --> To Ensure data security & Governance, data privacy
		
		
	ML Model Use Case :
	
		ML Model for Product Reviews :
		
		E_Commerce Application Eg. www.amazon.com
		
		Product Review:
		
		
	CICD Pipeline:										
		- SCM_Checkout 	==> VCS ?							 
		- Application Build
		- Create Application Artifacts 
		- Test 
		- Deploy 		
		

	Process/Lifecycle of Machine Learning Ops:::
	
		- Data Preparation 

		- Exploratory Data Analysis (EDA)
		
		- Feature Engineering
		
		- Model Training 
		
		- Model Validations 
		
		- Model Deployment 
		
		- Monitoring
		
	MLOps Use cases:


		- Healthcare 
		- Finance 
		- Retail Domain - E_Commerce
		- Manufacturing 
		
		
		
	How to Implement MLOps :::
	
	- Vesioning:
	
	Version Control System :
	
		- To Version Control the Changes 
		- To Track the Changes
		
	Types of VCS :
	
		- Centralized VCS:
		
		- Distributed VCS:
		
		GIT :
		
			- Git is an Open-Source Distributed VCS 
			- To Version Control the Changes 
			- To Track the Changes 
			- To perform Parallel Development using Branching Techniques
			
	
	

	Lab Requirements:
	
		- Create GitHub Account 
		
				- https://github.com/
		
		- Install Git Client on Local Machine 
			- Windows OS :
				- Git Bash
				- Git CMD 
				- Git GUI 
				
			- Linux/Mac OS : 
				- Terminal 


	Git - Client install on Local Machine 
	
	Github - Remote Repository Server
	
		

Next :::
		
		- Create AWS Free Tier Account 
		
		- Create DockerHub Account 
		
		- Install Visual Studio Code on Local Machine 
		
		- Install Python on Local Machine 
		


####################
Day2: 30th Nov. 2025
####################		
	
	MLOps Lifecycle :
	
	MLOps Usecase :
	
	Versioning :
		
		Version Control System using Git :
		
		Git Demo :
		
	ML Project and Packaging ML Model 
	
	CICD Pipeline:
	
		- SCM_Checkout 	==> VCS ?
		- Application Build
		- Create Application Artifacts 
		- Test 
		- Deploy 
		
		
	How to Implement MLOps :::

	
		- Data Engineers		-> Prepare the Source Datasets 
		
		- Data Scientists 		->  Explore the datasets, analyse, check the feasibility of datasets 
									Perform the training, tuning, and evaluate the Model
		
		- ML Engineers 			-> Deploy the ML Models to prod -> MLOps Strageies CICD, define best practises
									Should have MLOps Knowledge
		
		- Business Stakeholders -> Bring business values using this ML Models 
		
		- Data Security/Governance Team  --> To Ensure data security & Governance, data privacy
		
		
	ML Model Use Case :
	
		ML Model for Product Reviews :
		
		E_Commerce Application Eg. www.amazon.com
		
		Product Review:
		
		
	CICD Pipeline:										
		- SCM_Checkout 	==> VCS ?							 
		- Application Build
		- Create Application Artifacts 
		- Test 
		- Deploy 		
		

	Process/Lifecycle of Machine Learning Ops:::
	
		- Data Preparation 

		- Exploratory Data Analysis (EDA)
		
		- Feature Engineering
		
		- Model Training 
		
		- Model Validations 
		
		- Model Deployment 
		
		- Monitoring
		
	MLOps Use cases:


		- Healthcare 
		- Finance 
		- Retail Domain - E_Commerce
		- Manufacturing 
		
		
		
	How to Implement MLOps :::
	
	- Vesioning:
	
	Version Control System :
	
		- To Version Control the Changes 
		- To Track the Changes
		
	Types of VCS :
	
		- Centralized VCS:
		
		- Distributed VCS:
		
		GIT :
		
			- Git is an Open-Source Distributed VCS 
			- To Version Control the Changes 
			- To Track the Changes 
			- To perform Parallel Development using Branching Techniques
			
	
	

	Lab Requirements:
	
		- Create GitHub Account 
		
				- https://github.com/
		
		- Install Git Client on Local Machine 
		
				- https://git-scm.com/install/windows
				- https://git-scm.com/install/mac
				
			- Windows OS :
				- Git Bash
				- Git CMD 
				- Git GUI 
				
			- Linux/Mac OS : 
				- Terminal 
				
				git --version


	Git - Client install on Local Machine 
	
	Github - Remote Repository Server
	

	Working with Git :::
	
	
	Local Git Repository															Remote Git Repository
	
	
	

	
	Tracking/Versioning ?
	
	
	*.py 
	
	asdfasdf
	dfasdf
	asdfasdfasd
	----shion
	
	save myMLApp1.py		===> myMLApp1.py_v1.0
	
	save myMLApp1.py		===> myMLApp1.py_v1.1
	
	save myMLApp1.py		===> myMLApp1.py_v1.2

	save myMLApp1.py		===> myMLApp1.py_v1.3
	
	save myMLApp1.py		===> myMLApp1.py_v1.4		Latest version (Tag/Version_Number/Commit_ID)
	
	
	Git File Workflow :::

	Local Machine 																			Remote Server 
	
	Install git client 																			Github/AzRepo/gitlab
	
	Working_Directory 				Staging_Area 				Local_Repository				  Remote_Repsitory 
	
	 myMLApp1.py	 --------------> myMLApp1.py --------------> myMLApp1.py_v1.0 ----------------> myMLApp1.py_v1.0
						git add 					git commit						  git push
						
	 myMLApp1.py	 --------------> myMLApp1.py --------------> myMLApp1.py_v1.1 ----------------> myMLApp1.py_v1.1
						git add 					git commit						  git push
	
	


	Developer's Workload :
	
		Types of Projects:

			- Enhancement/Bugfix 
			- New Project 
			
	
	Git Cli Commands :
	
	
	git --version
	
	
	git clone 		:	To Clone/Copy the entire Remote Repository to Local Machine 
	
	git add 		:	To add the changes from Working Directory to Staging_Area
	
	git commit 		:	To commit the changes from Staging_Area to Local Repository
	
	git push 		:	To Push the Changes from Local Repository to Remote Repository
	
	git pull/Fetch 
		
		- Both git pull & git fetch are used to handle the incremental changes from Remote Repository
		
		- git Fetch :	
			To just check the incremental changes from remote repository. If any incremental changes exist, it will update the incremental changes only in the Local Repository. Not in the Working Directory.

		- git pull 	:
			Check the incremental changes from remote repository. If any incremental changes exist, it will update the incremental changes in the Local Repository as well in the Working Directory.

			As a best practise, always it is recommended to use git pull to keep the local repo in-sync with remote repository

				git pull = git fetch + git merge!
	
	git init 		:	To initialize a local git repository
						It will create a .git directory in the project workspace
						It will create a default branch called master/main
			
			
	Fork 			: 	To copy/backup a entire remote repository to another remote repository



	Work with Cli Commands :::
	
	Organize the WorkSpace :::
	
		cd d:

		mkdir MLProjects 
		
		cd MLProjects
		
		mkdir MLRepo1			# Workspace 
		
		mkdir MLRepo2
		
		cd MLRepo1
			git init 
	
	
		
	Local Machine 															
															
	
	Working_Directory 			Staging_Area 				Local_Repositor
	
	 file1.txt	 --------------> file1.txt --------------> file1.txt
					git add					git commit				
	
	
	
	git status			:	To Get the status of repository

	

	Git Configuration ::::
	
		As a best practise, using git global config to setup the user name and email after initializing the first local repository.
		
		Global Configuration:				# Applicable to all the Repositories
		
			git config --global user.name "LoksaiETA"
			
			git config --global user.email "loksai@asdf.com"

			git config --global --list
		
		Local Configuration:				# Applicable to a specific Repository
	
			git config user.name "LoksaiETA"
			
			git config user.email "loksai@asdf.com"	
			
			git config --list 
		
		
	git add ::::	
	
		git add file1.txt 

		git add file1.txt file2.txt
		
		git add *.doc 
		
		git add .							# . to select all files 
		


	git rm --cached 'file1.txt' 			# Unstage a file, the file changes will be back to working directory.
	
	git rm -f 'file1.txt' 					# Permantly Unstage and a delete a file from working directory.


Next :::
		
		
		- Git Reset / Revert 
		
		- Git Branches 
		
		- Git Branching Strategies
				
		- Git rebase / Merge
		
		- Git Remote Repositories Handling 		
		
		- Create AWS Free Tier Account 
		
		- Create DockerHub Account 
		
		- Install Visual Studio Code on Local Machine 
		
		- Install Python on Local Machine 
		
				
			
				
				
	
####################
Day3: 6th Dec. 2025
####################

	git 
	
	github 
	
	docker 
	
	dockerhub
	
	DVC 
	
	MLFLow 
	
	Kubernetes 
	
	KubeFLow 
	
	Prometheus 
	

	- git ignore :::
	
		- Is used ignore the files from tracking.
		- .gitignore is a hidden file, used to maintain the file names/patterns to ignore from tracking.
		- As a best practise, .gitignore should be the very first commit in any repository.
		
		
		
	Project_Workspace :		==> Push 
	
		- src 
			- *.py
			
		- test 
		
		- config_file
		
		- req.txt
		
		- app.prop 
		
		- db_credentials.json
		
		- Secret
		
		
	
	git log ::::
	
		git log 
		
		git log --oneline
		
	Undo the Commited Changes :::
	
	Undo/Revert the Committed Changes from Local Repository :::

	git reset :::
	
		git reset is used to undo the changes from the repository
		git reset will reset the commit point to the previous commit point 
		git reset will not create any new commit point for track purpose 
		git reset is not recommended in the shared repository 
	
		Syntax :		
			- git reset <reset_option> <previous_Commit_ID>
			
		Reset Options ::
		
			git reset --soft <previous_Commit_ID> :::
				- git reset will reset the HEAD pointer to the previous commit point.
				- It will take the changes back to staging area 
				- The Changes will be available in staging area and working directory					
		
			git reset --mixed <previous_Commit_ID> :::		# Default
				- git reset will reset the HEAD pointer to the previous commit point.
				- It will take the changes back to working directory 
				- The Changes will be available only in working directory			
			
			git reset --hard  <previous_Commit_ID> :::
				- git reset will reset the HEAD pointer to the previous commit point.
				- It will permanently delete the files from the system 
			
		
	ML Projects ::::
	
		Dataset  Source :
		
		
		ETL Process ::: 
		
			Extract 		==> 	for 30 mins  --> Raw Data ==> commited to a repository
			
			Transform 		==> 	Application Program to Process Data 	
			
			Load 			==> 	Saved to a Database 
			
							==>		Reset --hard pointing to initial commit (.gitignore)
							
	
	Git Revert :::
	
	git revert :::
	
		- Git Revert is same as git reset --hard option 		
		- git revert is used to undo a specific commit 		
		- git revert will create a new commit point for tracking purpose.		
		- git revert will maintain the commit history		
		- git revert is recommended in shared repositories
		
		Syntax ::
			git revert <specific_Commit_ID>		
			
	
	GIT Commit :::
	
	 - git commit 					# To Commit the changes from staging area to local repository	


			git commit -m "<Valid Commit_Message>"
			
			As a best practise :
			
			git commit -m "CR1001 - Message"
		
			git commit -m "REL-0126 - Payment Module updated"
							
							
	Git Branches ::::
	
		-> Git Branches are the logical copy of the Repository/Default Branch 
		
		-> Used to perform parallel development 
		
	
	Git Branching Strategies :::
	
		-> default branch - (master/main) ==> Used to maintain the production version of source.
		
		-> Git Branching Strategeies are used to maintain the integrity of the main/master Branch(Prod version)
	
	
	Repo1 :::
	
		master : cm1,cm2,c3 
		
			feature1_branch : cm1,cm2,c3
								cm1,cm2,c3,f1cm1,f1cm2,f1cm3,...................
	
	
	
	Scenario 1:
	
		Repo1 :::
	
			master : cm1,cm2,c3 			
						cm1,cm2,c3,cm4 						# 'cm4' is the changes merged from feature1_branch
			
				feature1_branch : cm1,cm2,c3
									cm1,cm2,c3,f1cm1,f1cm2,f1cm3
							
	
	Scenario 2:
	
		Repo1 :::
	
			master : cm1,cm2,c3
						cm1,cm2,c3,cm4 						# 'cm4' is the changes merged from Developer_Branch

				Developer_Branch : cm1,cm2,c3 
									cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
				
					feature1_branch : cm1,cm2,c3
										cm1,cm2,c3,f1cm1,f1cm2,f1cm3
					
					feature2_branch : cm1,cm2,c3
										cm1,cm2,c3,f2cm1,f2cm2,f2cm3

	Scenario 3:
	
		Repo1 :::
	
			master : cm1,cm2,c3
						cm1,cm2,c3,cm4 						# 'cm4' is the changes merged from Integration_Branch

				Integration_Branch : cm1,cm2,c3
										cm1,cm2,c3,Developer1_Changes,Developer2_Changes					
				
					Developer1_Branch : cm1,cm2,c3 
										cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
					
						feature1_branch : cm1,cm2,c3
											cm1,cm2,c3,f1cm1,f1cm2,f1cm3
						
						feature2_branch : cm1,cm2,c3
											cm1,cm2,c3,f2cm1,f2cm2,f2cm3			

					Developer2_Branch : cm1,cm2,c3 
										cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
					
						feature1_branch : cm1,cm2,c3
											cm1,cm2,c3,f1cm1,f1cm2,f1cm3
						
						feature2_branch : cm1,cm2,c3
											cm1,cm2,c3,f2cm1,f2cm2,f2cm3	
		
		
	Scenario 4:
	
		Repo1 :::
	
			master : cm1,cm2,c3
						cm1,cm2,c3,cm4 						# 'cm4' is the changes merged from Release_Branch
				
				Release_Branch : cm1,cm2,c3
									cm1,cm2,c3,Team1_Changes,Team2_Changes
					
						
					Integration_Branch1 : cm1,cm2,c3														Team-1
											cm1,cm2,c3,Developer1_Changes,Developer2_Changes					
					
						Developer1_Branch : cm1,cm2,c3 
											cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
						
							feature1_branch : cm1,cm2,c3
												cm1,cm2,c3,f1cm1,f1cm2,f1cm3
							
							feature2_branch : cm1,cm2,c3
												cm1,cm2,c3,f2cm1,f2cm2,f2cm3			

						Developer2_Branch : cm1,cm2,c3 
											cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
						
							feature1_branch : cm1,cm2,c3
												cm1,cm2,c3,f1cm1,f1cm2,f1cm3
							
							feature2_branch : cm1,cm2,c3
												cm1,cm2,c3,f2cm1,f2cm2,f2cm3	
			

					Integration_Branch2 : cm1,cm2,c3														Team-2
											cm1,cm2,c3,Developer1_Changes,Developer2_Changes					
					
						Developer1_Branch : cm1,cm2,c3 
											cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
						
							feature1_branch : cm1,cm2,c3
												cm1,cm2,c3,f1cm1,f1cm2,f1cm3
							
							feature2_branch : cm1,cm2,c3
												cm1,cm2,c3,f2cm1,f2cm2,f2cm3			

						Developer2_Branch : cm1,cm2,c3 
											cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
						
							feature1_branch : cm1,cm2,c3
												cm1,cm2,c3,f1cm1,f1cm2,f1cm3
							
							feature2_branch : cm1,cm2,c3
												cm1,cm2,c3,f2cm1,f2cm2,f2cm3	
			
			

	- Adhoc Requirements :::
	
		- Wotking with Hot-fix Branch :::
		
		
				Repo :
				
						master : cm1,cm2,cm3
						
									cm1,cm2,cm3,cm3.1			# 'cm3.1' is the changes merged from Hot-fix_Branch
							
								Hot-fix_Branch : cm1,cm2,cm3,hfcm1,hfcm2,hfcm3						
							
									Developer_Branch : cm1,cm2,c3 
														cm1,cm2,c3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
									
										feature1_branch : cm1,cm2,c3
															cm1,cm2,c3,f1cm1,f1cm2,f1cm3
										
										feature2_branch : cm1,cm2,c3
															cm1,cm2,c3,f2cm1,f2cm2,f2cm3		
			
			
			
	
	Working with Branch ::::
	
		Create Branch :
		
			git branch				# List the Branches 
			
			git branch feature1		# Create a branch 
			
			git switch -c feature1	# Create a branch and switch to new branch
			
			git switch feature1		# To Switch to a specific branch
			
			
			Repo:
			
				master : cm1 
				
					feature1 : cm1 
	
								cm1,f1cm1
								
								git switch master 				
								
								
								git merge feature1			# Should executed from the target branch
					
					
					feature2
					
					
					
					feature3
			
			

		
		
Next ::::

		Handle the Conflicts		

	
		
####################
Day4: 7th Dec. 2025
####################		
		
	Working with GIT Branches : - To Version Control the Source Code Changes 

	Working with Git Branches :


		- git branch 							# to get the list of branches 
		
		- git branch <new_branch_name>			# Create new_branch
		
		- git switch -c <new_branch_name>		# Create new_branch & Switch to new_branch.
		
		- git switch <exiting_branch_name>		# Switch to a branch
		
		
			Repo1: 
			
				- master - cm1,cm2,cm3

					git switch -c feature1				# Create new_branch & Switch to new_branch.
					
					feature1 :
							cm1,cm2,cm3,f1cm1,f1cm2 	# Perform incremental changes 
							
						git switch master 				# Switch to master branch
							
						git merge feature1				# Merge command should be executed from target branch
							
			
	GIT Merge Conflicts :

		- When more than one user/feature try to update same file at the same level, merge conflict occurs.
		
		
	Handle/Fix Merge Conflicts :

		- How to fix Merge Conflicts ?
		
			- Identify the file(s) causing the merge conflicts
			- Open the file and review the changes 
			- Upon review, decide which changes to be retained/delete from that file(s).
			- Update the file by removing the extra header/footer notes and contents.
			- Perform git add and git commit to fix the issue.
	
	On DevOps Perspective :

		- On DevOps Perspective, We never fix the Merge Conflict. Instead, we Prevent Merge Conflict.!
		
	
	GIT :::
		
GIT Rebase : 

		- Rebase is used to maintain linear commit history 
		- Rebase is used to keep the current branch in-sync with target branch 
		- Rebase can prevent merge-conflicts in the target branch 
		- As a best practise, it is always recommended to use rebase before merge
		
	Repo  :

		master : cm1,cm2,cm3 
					
					cm1,cm2,cm3,f1cm1,f1cm2,f1cm3 				# Upon Merging feature1 
					
					
					cm1,cm2,cm3,f2cm1,f2cm2,f2cm3,f1cm1,f1cm2,f1cm3		# Upon Merging feature2 without proper rebase
					
					cm1,cm2,cm3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3		# Expected Commit History! using rebase!
		
		
			feature1 : cm1,cm2,cm3 
			
						cm1,cm2,cm3,f1cm1,f1cm2,f1cm3 
						
						git switch master 
						
						git merge feature1 
						
			
			feature2 : cm1,cm2,cm3 
				
						cm1,cm2,cm3,f2cm1,f2cm2,f2cm3 
						
						git rebase master 
						
						cm1,cm2,cm3,f1cm1,f1cm2,f1cm3,f2cm1,f2cm2,f2cm3
						
						
						git switch master 
						
						git merge feature2		
		
		
	- merge 
	
	- rebase 	--> Within the Repo. between the branches 
	
	- pull 		--> Used between the Remote & Local 
		
		

	- GIT SQUASH :::
	
		- It is a merge Option. Is used to combine more than one commits into single commit point.
		
		
		Repo1: 
		
			master: cm1,cm2 
			
						cm1,cm2,f1cm1,2,3,4,5,6,7,8,................f1cm100   # Upon Merge without squash option

						cm1,cm2,cm3 					# Cm3 is combined changes from feature1
			
				feature1 : cm1,cm2 
				
							cm1,cm2,f1cm1,2,3,4,5,6,7,8,................f1cm100
							
							
					git switch master 
					
					# git merge feature1 
					
					git merge --squash feature1 
			
					git commit -m "Combined Commit from feature1"
	

	- GIT STASH :::
	
		- Used to save the uncommited changes to a temporary area.
		
		
	git stash save "save_message"	# Save an entry in stash list
	
	git stash list 					# Get the Stash list.
	
	git stash apply 				# To apply the latest entry from the stash list back to staging area. 	
	git stash apply stash@{2}		# To apply a specific entry from the stash list back to staging area
	
	git stash drop 					# To delete the latest entry from the stash list.
	git stash drop stash@{2}		# To delete a specific entry from the stash list.
	
	git stash pop 					
		# To apply the latest entry from the stash list back to staging area as well as delete that latest entry from the stash list.								
									
	git stash pop stash@{1}	
		# To apply a specific entry from the stash list back to staging area as well as delete that specific entry from the stash list.
									
	git stash clear 				# To clean-up the stash list 
		
	
		
	GitHub Remote Repository Management :::
	
		- Create Remote GitHub Repositories:
		
		- Create Branches in Github Repositories
		
		- Merge the changes to target branch -- using pull request 
		
	
	Developers' Workload :
	
		--> Enhancement/Bugfix 
		
		--> New Project 
		
		

	git clone 		:	To Clone/Copy the entire Remote Repository to Local Machine 
	
	git add 		:	To add the changes from Working Directory to Staging_Area
	
	git commit 		:	To commit the changes from Staging_Area to Local Repository
	
	git push 		:	To Push the Changes from Local Repository to Remote Repository
	
	git pull/Fetch 
		
		- Both git pull & git fetch are used to handle the incremental changes from Remote Repository
		
		- git Fetch :	
			To just check the incremental changes from remote repository. If any incremental changes exist, it will update the incremental changes only in the Local Repository. Not in the Working Directory.

		- git pull 	:
			Check the incremental changes from remote repository. If any incremental changes exist, it will update the incremental changes in the Local Repository as well in the Working Directory.

			As a best practise, always it is recommended to use git pull to keep the local repo in-sync with remote repository

				git pull = git fetch + git merge!
	



	GIT Clone :::
	
		- git clone https://github.com/Edu-MLOps-BA1125/TestRepo1.git

	
		- git remote -v		# To get the list of remote repositories linked to a local repository
		
		
		
		
		- Token Based Authentication :::
		
			- PAT ==> ghsdfsdfasdfasdfM0v
		
		

Next :::
		
	Docker & DockerHub 
	
		- Manage the Container Images 
		
		- DockerHub to Version Control the Container Images 
		
	DVC 

	MLFlow 
	
	KubeFLow 



#####################
Day-5: 13th Dec. 2025
#####################	

	- Packaging the ML Project :::
	
	- Docker ->
		
	- Containerization :::
		- Docker & DockerHub 
		
	- DVC => Version Control the ML Dataset 
	
	- AWS S3 Buckets ==> To Manage the ML Dataset ==> git & github  ===> DVC & AWS S3.
	
	- ML Project 
		- Python -> Streamlit, Flask, Fast API		
		
	- github actions -> CI/CD Pipeline 
	
		- MLFlow 
		
	- End-to-End CICD MLOps Pipeline using github-actions  
	
	- AWS Sagemaker
		
		
Containerization using Docker ::::	To Package the ML Model!

	- Containerization ::::
	
		- Is a process of packaging the Application along with the dependencies.
		
	- Docker is Container Engine :
		- Is to Create/Manage the Container Images and Containers
		
	Terminologies :::
	
		- Docker Engine :	Is to Create/Manage the Container Images and Containers
		
		- Container Images :	
				- Is a static File, that define the dependencies to package the Application
				- It is Non-Executable
				- It composed of many Layer of Instruction
				
		- Container :
				- Is an Executable unit of Container Images.
		
		- Container Registry :
				- Used to Version the Container Images. 
				- DockerHub*** / ECR / ACR / GCR 
				- Create a DockerHub Account
		
		- Container Repository :
				- Sub-set of Container Registry
		
		Dev Environment:		
		
			*.java 		
					
			github:						Artifactory Library			Container Registry
											Jfrog						DockerHub 

				*.java_V1.0		==> 	*.war_v1.0				==> 	mywebapp_img:v1.0
				*.java_V1.1		==> 	*.war_v1.1              ==> 	mywebapp_img:v1.1
				*.java_V1.2		==> 	*.war_v1.2              ==> 	mywebapp_img:v1.2
				*.java_V1.3		==> 	*.war_v1.3              ==> 	mywebapp_img:v1.3
				
			DVC & AWS-S3/Azure Disk/G Volume
			
				==> Maintain the Version of ML Datasets.
		
	Virtual Machine :::
	
		- VMs are called as Hardware Level Virtualization
		- VMs are created using Hypervisor
		- VMs are used to Execute the Operating System 
		- VMs will continue to execute, even if there no Active Task/Application.
		- VMs consume more space and time to startup the Application
	
	Container :::	

		- Containers are called as OS Level Virtualization
		- Containers are created using Container Engine like Docker
		- Containers are used to Execute the Task/Application. Not Operating System 
		- Containers will go to EXIT State, if there no Active Task/Application.
		- Containers consume less space and time to startup the Application	
		
		- Containers run in it's own dedicated/Isolated Address space.
		
		- Containers uses the properties of Underlying Operating System
		
			Kernel :
				- Namespace 
				- Control Group
		
	Work with Docker :::
	
		1. Explore DockerHub Images 
		
		2. Work with Container Images & Containers using Docker CLI Commands. 
		
		3. Create Container Images 
		
		4. Publish to Container Registry 
		

		- Create DockerHub Account 			# https://hub.docker.com/
		
		- Install Docker Engine on Build Environment
		
			- Use AWS EC2 instance - Create Ubuntu Machine 
			
			- Login to AWS EC2 Instance using SSH Agent 
			
			- Install Docker Engine
		

	- Access AWS Cloud Platform :::
	
	Access AWS Cloud ! 
	
		- Create AWS Free Tier Account !		
		
			# https://aws.amazon.com/console/
			# https://docs.aws.amazon.com/accounts/latest/reference/manage-acct-creating.html
				
			==> Active Email_ID 
			==> Active Mobile Number 
			==> Active Credit/Debit Card - Enabled with International Transactions 
			
	Upon Creating an Account :

	Login to AWS Console:
	
		- Email_ID 			# AWS root user ID.
		
		- Login as a root user.
	
			- EMail_ID 
			- Password 
			
	
	- Launch an EC2 Instance - VM 
	
	
	- Connect & Loginc to EC2 Instance :::
	
		- EC2 Instance Connect Option 		# Thru Web Browser 
		
		- SSH Agent 		# Windows users 
			- MobaXterm 	# https://mobaxterm.mobatek.net/download-home-edition.html
			- Putty 
			
		- Terminal			# Mac/Linux User
		
	
	- Install Docker Engine & work with Docker cli commands :::
	
	
	sudo -i 
	
	apt update -y 
	
	apt install docker.io -y 
	
	- Work with Docker Cli Commands :::

	docker --version 
	
	docker images 		# To get list of Docker Images in the Local Machine
	
	docker ps			# To get the list of Active Containers 
	
	docker ps -a 		# To get the list of of Active & Inactive Container 
	
	Use DockerHub and working with existing Images :::
	
	docker pull <image_name>			# Download an Image from Container Registry
	
	docker pull <image_name>:<Version_No>
	
	docker run <image_name>				# Create Container & Execute
	
	Container Execution Modes :
	
		1. Foreground Mode/Attached Mode :			# Default Mode 
		
			- docker run ubuntu sleep 20 
		
		2. Background Mode/Detached Mode :

			- docker run -d ubuntu sleep 20 
			
		3. Interactive Mode :
		
			- docker run -it ubuntu bash 
			
				- exit 					
				
	
	docker start <Container_ID>
	
	docker exec -it <Container_ID> bash 			# Login to the Running Container
	
		- exit										# Just Log-off  		
	
	docker stop <Container_ID>
	
	
	- Create an Application Images
	
	- Push that Image to Container Registry 
	
	- Pull to any target Machine and Create Container
	
	
	How to Create a Container Image :::
	
	- Using Docker Build Command :
		- Pre-Requisites : Dockerfile 
	
		Syntax :
			docker build -t <Container_Repository_Name>/New_Image_Name:<v1.0> .	# "." denotes the Dockerfile ref.
		
		- docker build -t loksaieta/AppImg:v1.0 .
	
	- Create Dockerfile :::
	
		FROM ubuntu 
		RUN apt update -y 
		RUN apt install git -y 
		RUN apt install nginx -y 
		
	
	Dockerfile Instructions :
	
			FROM 									# To Identify the Base Image
			RUN 									# To run the package manager
			COPY 									# To Copy the file from host volume to container volume
			CP 										# To Copy the file within the container volumes 
			ADD 									# To Copy the file from Host Volume as well as from URL
			ENV 									# To define the Environment Variable 
			ARG 									# To pass Arguements to the Steps in Dockerfile 
			EXPOSE									# To Define the Container Port 
			WORKDIR									# To set the current working directory within the Container 
			CMD										# To set the default start-up command to the container 
													  This Command can be changed at run-time.
			ENTRYPOINT								# To set the default start-up command to the container 
													  This Command cannot be changed at run-time.	
	
	Docker Build :::


Next :::
	
	Docker Push :::
	
		- To Publish the Image to Container Registry :::
		
		-> DockerHub Login User Name 
		
		-> DockerHub User's PAT - Access Token 
		
			- Create Access Token On DockerHub 
		
	- Port Mapping : Access the Container Application 


	
######################
Day-6: 14th Dec. 2025
######################	


	-> Publish the new Version of Docker Images to DockerHub :::
	
		DockerHub :
		
			Login to DockerHub using Docker Cli Command :
			
				docker login	:
				
					DockerHub Login ID 			docker login -u loksaieta
					DockerHub Access Token		sdfsdfsdfsdfs
					
					
					
			docker push <Container_Repository_Name>/New_Image_Name:<v1.0>
	
	
	-> Port Mapping!
	
		- Is used to expose the Application running inside the container to Internet.
		
		
			Tomcat WAS: Java Web Projects 		8080
			
			Streamlit WAS : ML Projects 		8501
			
			
			docker run -p 8080:8501 <image_name> 
			
			docker run -it <image_name> 
			
			
			-p <host_port>:<container_port>
			
			
			
	



		-> Web Application :
		
			ML Model:
			
				chatbot 
				
				Amazon Review 
				
				
				
			github:						Artifactory Library			Container Registry
											Jfrog						DockerHub 

				*.java_V1.0		==> 	*.war_v1.0				==> 	mywebapp_img:v1.0
				*.java_V1.1		==> 	*.war_v1.1              ==> 	mywebapp_img:v1.1
				*.java_V1.2		==> 	*.war_v1.2              ==> 	mywebapp_img:v1.2
				*.java_V1.3		==> 	*.war_v1.3              ==> 	mywebapp_img:v1.3
				
	DVC :::	 https://dvc.org/
	
	
	DVC & AWS-S3/Azure Disk/G Volume
			
		==> Maintain the Version of ML Datasets.
		
		--> Virtual Environment :::
		
				To create an isolated platform to work on ML Projects 
				
				1. MLProject - DVC ==> S3 Bucket   # Create Venv1	==> Proj1/			
					
				
				2. MLProject - DVC ==> GDrive 		# Create Venv2	==> Proj2/
				
				
				git (Version Control Source Code)		<===>			dvc (Version Control ML Datasets)
				
				github(Remote-Repo)										S3/Azure Disk/GDrive/Persistant Volumes
				
		--> Install in Dev/Build Environment :::
		
		
	Working with DVC :::
	
	Pre-Requisites :
	
		--> AWS EC2 Ubuntu Instance :::
		
		--> AWS S3 Bucket ::: 
		
		--> AWS Cli => AWS Access & Secret Key to Enable Authentication / IAM Role attached to EC2 Instance
		
		--> 
		
	1. Login to AWS Console & Create S3 Bucket :::			Bucket ::: loksai-edu-mlproject1
	
		
{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "Statement1",
            "Effect": "Allow",
            "Principal": "*",
            "Action": "s3:*",
            "Resource": "arn:aws:s3:::loksai-edu-mlproject1/*"
        }
    ]
}
		
		IAM Role: loksai-edu-mlproject1-s3role
		
		Launch EC2 Instance attach IAM Role
		
		
Login to EC2 Instance :::::

sudo -i 

1️⃣ Update system

sudo apt update -y


# To Setup AWS CLI ::::


sudo apt install -y unzip curl

2️⃣ Download AWS CLI v2
curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip"

3️⃣ Unzip installer
unzip awscliv2.zip

4️⃣ Install AWS CLI
sudo ./aws/install

5️⃣ Verify Installation
aws --version


Expected output:

aws-cli/2.x.x Python/3.x Linux/x86_64


✔ AWS CLI installed successfully


# To Setup EC2 Instance with AWS ::::




1.2 Install Required Tools

sudo apt update -y 

sudo apt install -y \
  python3.12 python3.12-venv python3-pip
  
  

1.3 Configure AWS CLI

aws configure


Provide:

Access Key	:	

Secret Key	:

Region (e.g. ap-south-1)

Output format: json

2️⃣ ENTER PROJECT & CREATE VENV


python3.12 -m venv venv

source venv/bin/activate


pip install dvc_s3									# Install DVC-s3 



mkdir dir1 

git init 

dvc init 

dvc remote add -d myremote s3://loksai-edu-mlproject1



dvc push 


deactivate 

dvc pipeline :

	dvc repro 		# Reproduce 

	1. DVC is used to manage the Input Dataset for Model Training 
	
	2. dvc repro 		# Reproduce  ==> Train the Model 

	3. Trained Model ==> ML Artifacts ==> model.joblib / pickle 
	
	4. Used DVC to maintain the Artifacts - S3 
	
	5. Use that artifacts and dependencies to create Dockerfile and create Docker Image 
	
	
Next :	
	MLFLow :::
	
	Kubeflow :::
	
	ML Application Tools -> Web Applications/API --> Streamlit 
	

ML_Model
	-> ML Artifacts :::		Trained Model 


MLOps Project ::::

	- Create an ML Model using 
	
		-> Streamlit, Fast-API,
		
	1. Simple Web Application :
	
	
	2. Prod Level Application :
	
	
	DevOps / MLOps Team / LLMOps :
	
	Roles & Responsibilities ::
	
	- Detailed Assessment of the requirement 
	
		- Interact with all the ML Teams --> DS,DE, ML Engg, 
		
		Tools to be used :
		
		MLOps Stack!
		
			- Source Code :		Application Source Code - API, UI, Pipeline(Train the Model)
			
			- Package 	  :  	Docker & DockerHub 
			
			- Artifacts   :		Trained Model --> joblib/pickle - model.pkl  
			
			- Depeloyment  :	DockerHub 	# get the Container Images 
								AWS S3 	  	# get the Model & Data
								Kubernetes	# To Orchestrate the Containers 
								Kubernetes Cluster is the Target Environment
								
								
		Kubernetes ::::
		
			- It is a Open-Source Container Orchestration Tool 
			- Kubernetes is used to Deploy any type of Containers.
			- It is used to ensure high availability of the Applications/services running thru Containers.
			- Used to Ensure High Availability of Containers by creating Replicas of Containers.
			- It supports Auto-Scaling & Load Balancing.

			- Self-healing Capability!
			
			- Using this we can achieve Continuous Deployment


			
		Kubernetes Architecture :::
			
			Architecture Components :::

		- Kubernetes Architecture components :::
		
			API_Server 				--> # Acts as an interface to the kubernetes 
			
			ETCD 					--> # Single point of Source for Kubernetes Components 
			
			Scheduler				--> # To identify the Healthy Node for Deployments
			
			Controller Manager 		--> # To run the pods in its desired state 
			
			
			Kubelet 				--> # Is a Kubernetes Agent used to Create & Deploy the Pods
			
			KubeProxy				--> # Is used to enable pod networking by create Pod IP Address			
			
			CRI - Container RunTime Interface (Container-D)
									--> # It is used identify the Image from Container Registry for deployment							
			
			
			
		- Terminologies :::
		
			- Container Images :
				- It is a static file used to define the properties and dependencies of application
				- These are non-executable 
				- It is composed of various layers of instruction created using Dockerfile.
						
			- Container :
			
				- Container is an executable entity of Container Image
			
			- Container Registry :
				- Container Registry is used to Store/Manage the Container Images.
				- DockerHub is a default Container Registry for Docker Container Engine.
				https://hub.docker.com/
				
				
			- Container Repository :				Public/Private 
				- Sub-set of Container Registry 
				
				

			- Kubectl 						# It is Command Line Utility to interact with Kubernetes Master 
			
			- Kubernetes Cluster 
											# Is a collection of Worker Nodes. 
							
			- Pod 							# Atomic Unit of Scheduling 			
			
			- Kubernetes_Master				# Used to Create and Schedule the Deployments to Kubernetes_WorkNodes
			
			- Kubernetes_WorkNodes			# Target Servers 
			
			
			--> Deploy 5 Replicas of Pod:
					==> Can be distibuted to all the available nodes by kubernetes
					
			
	- To Setup the Kubernetes Architecture :::

			Kubernetes_Master 
				Kubernetes_WorkerNode1
				Kubernetes_WorkerNode2
				
				
	- Minikube, K3s		# Learning basic kubectl commands 
						# Single Node Cluster 		

						# https://docs.k3s.io/installation
						
						
		Using K3s :
		
			Use AWS Ubuntu - 22.04V AMI 
			
			- sudo -i 
			
			- apt update -y  
			
			- curl -sfL https://get.k3s.io | sh -
			
			- sudo systemctl status k3s
			
			- kubectl get nodes 
			
			- sudo hostnamectl set-hostname "k3s-master-node"
			
			- exec bash
			
	
	
	- Working with Kubernetes Deployment Controller Object :::
	
	vi deploy-ui.yaml 			# Used to define the Pods for UI	- Streamlit UI
	
	vi nodeport-svc-ui.yaml		# Nodeport is used to expose the pods to internet


	vi deploy-api.yaml 			# Used to define the Pods for APIs 	- Fast API 
			
	vi nodeport-svc-api.yaml	# Nodeport is used to expose the pods to internet
	


	Create a build server :
	
	
	sudo -i 
	
	source venv/bin/activate
	
	pip install -r requirements.txt
	
	cd Proj_directory 
	
	> 
	
git init 

dvc init 

dvc remote add -d s3remote s3://edu-loksai-mlproj-bkt1

git add .dvc/config

git commit -m "Configure DVC S3 remote"

vi data/raw/reviews.csv				# Add sample Records			

dvc add data/raw/reviews.csv		# Add data/raw/reviews.csv to git after every dvc add

git add data/raw/reviews.csv.dvc data/raw/.gitignore

git commit -m "Added DVC Data-v1"

dvc repro


git add data/processed/.gitignore models/.gitignore dvc.lock			# Add dvc.lock to git after every dvc repro

git commit -m "repro-Processed data-v1"

dvc push 							# Push repro-Processed data 

dvc status							# Status should be: Data and pipelines are up to date.
	
			

	
	

######################
Day-7: 20th Dec. 2025
######################	

	Work with MLOps Projects ::::
	
	List of Tools/Services :
	
	- Monitoring the ML Model :
	
		Open-Source Tools :
		
			1. MLFLow :::	Tracking & Experiment 
			
			GUI Based Monitoring 
			
				- To ensure the expected level of accuracy ==> 95% Accuracy depends on the data!	
		
			2. Prometheus & Grafana
			
				- Infra-Structure Monitoring & the Kubernetes Pods Monitoring

	
		Managed Services from AWS --> Paid Service.
		
			1. AWS Sagemaker ==> Use cases/Application ==> 8GB RAM 
			
	
	- Application Web Services :::
	
		- Fast API 		==> Acts as a Backend 
		
		- Streamlit		==> User Interface.
		
		
	- Packaging the Model :::
	
		- Docker & DockerHub 
		- ECS/ECR 
		
			- Application(Model Scripts) & its dependencies
			
			- Do not package Data and Model 
			
			- Create dedicated Containers for API and UI
			
	- Container Orchestration Tool :
	 
		- Kubernetes Open-Source 
		- EKS/AKS/GKE
		
	- github-actions CICD Pipeline 
	
	

MLOps Project ::::

	- Create an ML Model using 
	
		-> Streamlit, Fast-API,
		
	1. Simple Web Application :
	
	
	2. Prod Level Application :
	
	
	DevOps / MLOps Team / LLMOps :
	
	Roles & Responsibilities ::
	
	- Detailed Assessment of the requirement 
	
		- Interact with all the ML Teams --> DS,DE, ML Engg, 
		
		Tools to be used :
		
		MLOps Stack!
		
			- Source Code :		Application Source Code - API, UI, Pipeline(Train the Model)
			
			- Package 	  :  	Docker & DockerHub 
			
			- Artifacts   :		Trained Model --> joblib/pickle - model.pkl  
			
			- Depeloyment  :	DockerHub 	# get the Container Images 
								AWS S3 	  	# get the Model & Data
								Kubernetes	# To Orchestrate the Containers 
								Kubernetes Cluster is the Target Environment
								
								
		Kubernetes ::::
		
			- It is a Open-Source Container Orchestration Tool 
			- Kubernetes is used to Deploy any type of Containers.
			- It is used to ensure high availability of the Applications/services running thru Containers.
			- Used to Ensure High Availability of Containers by creating Replicas of Containers.
			- It supports Auto-Scaling & Load Balancing.

			- Self-healing Capability!
			
			- Using this we can achieve Continuous Deployment


			
		Kubernetes Architecture :::
			
			Architecture Components :::

		- Kubernetes Architecture components :::
		
			API_Server 				--> # Acts as an interface to the kubernetes 
			
			ETCD 					--> # Single point of Source for Kubernetes Components 
			
			Scheduler				--> # To identify the Healthy Node for Deployments
			
			Controller Manager 		--> # To run the pods in its desired state 
			
			
			Kubelet 				--> # Is a Kubernetes Agent used to Create & Deploy the Pods
			
			KubeProxy				--> # Is used to enable pod networking by create Pod IP Address			
			
			CRI - Container RunTime Interface (Container-D)
									--> # It is used identify the Image from Container Registry for deployment							
			
			
			
		- Terminologies :::
		
			- Container Images :
				- It is a static file used to define the properties and dependencies of application
				- These are non-executable 
				- It is composed of various layers of instruction created using Dockerfile.
						
			- Container :
			
				- Container is an executable entity of Container Image
			
			- Container Registry :
				- Container Registry is used to Store/Manage the Container Images.
				- DockerHub is a default Container Registry for Docker Container Engine.
				https://hub.docker.com/
				
				
			- Container Repository :				Public/Private 
				- Sub-set of Container Registry 
				
				

			- Kubectl 						# It is Command Line Utility to interact with Kubernetes Master 
			
			- Kubernetes Cluster 
											# Is a collection of Worker Nodes. 
							
			- Pod 							# Atomic Unit of Scheduling 			
			
			- Kubernetes_Master				# Used to Create and Schedule the Deployments to Kubernetes_WorkNodes
			
			- Kubernetes_WorkNodes			# Target Servers 
			
			
			--> Deploy 5 Replicas of Pod:
					==> Can be distibuted to all the available nodes by kubernetes
					
			
	- To Setup the Kubernetes Architecture :::

			Kubernetes_Master 
				Kubernetes_WorkerNode1
				Kubernetes_WorkerNode2
				
	- Kubeadm 			# Multi-Node Cluster with full capabilities 
						# Highly recommended for Prod Implementation
						
						# https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/install-kubeadm/
						
	- Minikube, K3s		# Learning basic kubectl commands 
						# Single Node Cluster 		

						# https://docs.k3s.io/installation
						
						
		Using K3s :
		
			Use AWS Ubuntu - 22.04V AMI 
			
			- sudo -i 
			
			- apt update -y  
			
			- curl -sfL https://get.k3s.io | sh -
			
			- sudo systemctl status k3s
			
			- kubectl get nodes 
			
	
	
	- Working with Kubernetes Deployment Controller Object :::
	
	vi deploy-ui.yaml 			# Used to define the Pods for UI	- Streamlit UI
	
	vi nodeport-svc-ui.yaml		# Nodeport is used to expose the pods to internet


	vi deploy-api.yaml 			# Used to define the Pods for APIs 	- Fast API 
			
	vi nodeport-svc-api.yaml	# Nodeport is used to expose the pods to internet
	


	Create a build server :
	
	
	sudo -i 
	
	source venv/bin/activate
	
	pip install -r requirements.txt
	
	cd Proj_directory 
	
	> 
	
git init 

dvc init 

dvc remote add -d s3remote s3://edu-loksai-mlproj-bkt1

git add .dvc/config

git commit -m "Configure DVC S3 remote"

vi data/raw/reviews.csv				# Add sample Records as shown below		

dvc add data/raw/reviews.csv		# Add data/raw/reviews.csv to git after every dvc add

git add data/raw/reviews.csv.dvc data/raw/.gitignore

git commit -m "Added DVC Data-v1"

dvc repro


git add data/processed/.gitignore models/.gitignore dvc.lock			# Add dvc.lock to git after every dvc repro

git commit -m "repro-Processed data-v1"

dvc push 							# Push repro-Processed data 

dvc status							# Status should be: Data and pipelines are up to date.
	
	
######################
Day-8: 21st Dec. 2025
######################	


	Model Management :

	Work with the Trained Model :::
		- mlflow 
		- Model API 
		- Model UI 

# Stop any running MLflow process
pkill -f mlflow

# Remove ALL server overrides (VERY IMPORTANT)
unset MLFLOW_UI_SERVER
unset MLFLOW_GUNICORN_OPTS
unset MLFLOW_WAITRESS_OPTS

# Start MLflow UI (ONLY correct command)
mlflow ui --host 0.0.0.0 --port 5000 --allowed-hosts '*'

# Open in browser
http://<EC2_PUBLIC_IP>:5000


### OPTIONAL: MODEL VALIDATION
# Run API locally
uvicorn api.main:app --host 0.0.0.0 --port 8000

# Open in browser
http://<EC2_PUBLIC_IP>:8000

Eg.: http://13.201.128.37:8000/docs

### Run Streamlit
streamlit run ui/app.py --server.address 0.0.0.0			### Confirm predictions change with new data.

# Open in browser
http://<EC2_PUBLIC_IP>:8501


	Upload the Model to s3 
	
	
	
	Containerize the ML Project - API & UI 
	
		- Docker Image -> ML Project - API & UI 
						-> Should not have any Trained Model
							- This Trained Model should received at runtime
	
	
	use the Docker.api & Docker.ui   # Dockerfiles to Build Image.
	
		docker build -f docker/Dockerfile.api -t loksaieta/api830 . 
	
		docker build -f docker/Dockerfile.ui -t loksaieta/ui830 . 

	
	Publish the ML Project - API & UI to DockerHub 
	
		docker login -u <dockerhub_login_id>
		
		Password: <DockerHub_PAT>
		
		docker push loksaieta/api830
	
		docker push loksaieta/ui830
		
		
		
	
	Publish the ML Project to Github Repository 			# https://github.com/Edu-MLOps-BA1125/Loksai-MLProject-e2e
	
	
	
	
	
	MLOps Process :::
	
	Create CI CICD Pipeline :
	
		Build Orchestration Tool :
		
			- Jenkins / Az Pipelines / AWS CodePipelines/ gitlab-ci / github-actions workflow 
			
			
			Build_Environment :
			- Dynamic Build GitHub Server 
			
				-  Source Code Check out 
				-  Perform Build Docker / Image Build  
				-  Publish the Build Artifacts / Image to Docker Registry
			
			Target Environment : -> QA/STAGE/PROD 
			- Self Hosted Agent 
			
				-  Deploy to Target Environment 
				
			Pre-Requisites to work on Github Action workflow :

				--> Eg.: DockerHub Credentials 
				
				1. Create Secret_Variables - DockerHub Username & PAT 
				
				
	
		Github-Actions :::::
	
	Automate the Model Creation and Containerization using CICD Workflow - github-actions workflow.
	
	Setup-self-hosted Agent on github
	
	Deploy using Kubernetes 
	
		1. Launch a Target Server / Kubernetes Cluster & Attach IAM Role for S3.
		2. Open all the required ports 
		3. Login and Install the required packages
				- Docker / K3s/Minikube/Kubeadm 
		
				- Install unzip utility / curl 
				
				- Install aws-cli 
				
				- aws configure 
					- AWS: Access Token & Secret Key
					
		4. Setup the Credentials:
			- Create linux user - devopsadminuser1
				- valid Credentials - 
					 Recommended to have ssh-key - public_Key & Private Keys 

#Add User : 

useradd devopsadmin -s /bin/bash -m -d /home/devopsadmin

su - devopsadmin

#ssh-keygen

#for Ubuntu ::
#ssh-keygen -t rsa -b 2048 -m PEM								#ubuntu 20.04

ssh-keygen -t ecdsa -b 521										#ubuntu 22.04 or higher version of ubuntu				


ls ~/.ssh 

#You should see following two files:

#id_ecdsa - private key
#id_ecdsa.pub - public


#cat id_rsa.pub > authorized_keys

cat id_ecdsa.pub > authorized_keys

chmod 600 /home/devopsadmin1/.ssh/*

					 
				- Enable user access to kubernetes Master 
				
					- Create .kube/config under user's home dir. 
					- Use this user's kube/config file content as token for Authentication on github runner.
					
					
		5. Setup github self-hosted action-runner
		
			- Goto the Repo level setting 
			- Actions -> Runner 
			- Create Runner based on the OS & follow the instructions. 
					
			ssh conn 


Next: e2e deployment using k8s 
		Monitor - mlflow - 
			Prometheus & Grafana for Infra-Structure Monitor
		
		Use Case for AWS Sagemaker 


github-actions workflow ::: # cicd pipeline

name: CI-CD Pipeline

on:
  push:
    branches:
      - main

jobs:
  # =============================
  # JOB 1: Build & Push Images
  # =============================
  build:
    name: Build & Push to DockerHub
    runs-on: ubuntu-latest

    steps:
    - name: Checkout code
      uses: actions/checkout@v4

    # Login to DockerHub
    - name: Login to DockerHub
      uses: docker/login-action@v2
      with:
        username: ${{ secrets.DOCKER_USERNAME }}
        password: ${{ secrets.DOCKER_PASSWORD }}

    # Build API image
    - name: Build API Docker image
      run: |
        docker build -t ${{ secrets.DOCKER_USERNAME }}/loksai-mlops830api:latest -f docker/Dockerfile.api .

    # Build UI image
    - name: Build UI Docker image
      run: |
        docker build -t ${{ secrets.DOCKER_USERNAME }}/loksai-mlops830ui:latest -f docker/Dockerfile.ui .

    # Push API image
    - name: Push API image
      run: |
        docker push ${{ secrets.DOCKER_USERNAME }}/loksai-mlops830api:latest

    # Push UI image
    - name: Push UI image
      run: |
        docker push ${{ secrets.DOCKER_USERNAME }}/loksai-mlops830ui:latest

  # =============================
  # JOB 2: Deploy to Kubernetes
  # =============================
  deploy:  
    name: Deploy to Kubernetes
    runs-on: self-hosted
    needs: build

    steps:
    - name: Checkout code
      uses: actions/checkout@v4

    - name: Setup kubectl
      run: |
        mkdir -p ~/.kube
        echo "${{ secrets.KUBECONFIG_DATA }}" > ~/.kube/config


    - name: Deploy to Kubernetes
      run: |
        kubectl apply -f k8s/api-deployment.yaml
        kubectl apply -f k8s/ui-deployment.yaml



	

# Create data/raw/review.csv 


review_text,sentiment
I love this product it works perfectly,positive
Absolutely fantastic quality and service,positive
Very happy with the purchase,positive
Great experience would buy again,positive
Excellent product highly recommended,positive
Amazing performance and value,positive
Satisfied with quality and delivery,positive
Works exactly as advertised,positive
Customer support was very helpful,positive
This is the best purchase I have made,positive

This product is okay not great not bad,neutral
Average quality nothing special,neutral
It works but has some issues,neutral
Neither good nor bad experience,neutral
Product meets basic expectations,neutral
Decent but could be improved,neutral
Acceptable quality for the price,neutral
Nothing impressive but usable,neutral
Fine for occasional use,neutral
Okayish experience overall,neutral

Worst product I have ever bought,negative
Very disappointed with the quality,negative
Terrible experience waste of money,negative
Does not work as expected at all,negative
Extremely poor build quality,negative
Bad experience will not recommend,negative
Product stopped working in two days,negative
Customer service was terrible,negative
Totally useless and overpriced,negative
Highly dissatisfied with this product,negative

I am happy with the fast delivery,positive
Packaging was good and product works well,positive
Average packaging but product is decent,neutral
Late delivery and damaged box,negative
The item arrived broken,negative
Very smooth and reliable performance,positive
Not worth the price paid,negative
Okay product but slow response,neutral
Loved the design and usability,positive
Disappointed with the overall experience,negative
	


	
	
#####################
Day-9: 3rd Jan. 2026
#####################

	- Continuous Monitoring :
	
		- Various Tools and Services for ML Monitoring
			- MLflow 
			- Prometheus & Grafana 
				- Prometheus 	==> Monitoring Tool 
				- Grafana 		==> Visualization Tool 
			
				-> K3s - Kubernetes pods, Nodes, API Monitoring 
				
				
		- Monitoring Tools Architecture :::
		

			
			
	- Kubernetes Cluster
			
		- To Setup Prometheus & Grafana using Helm
		
		zero-down-time
		
		
		K3s Installation & Configuration ::::
		
		
		Using Helm - to To Setup Prometheus & Grafana
		
			- Helm is a package manager for Kubernetes
		
		
		
	- Kubeflow for Kubernetes :::
	
		

	- git,github,dvc,s3,iam,mlflow,docker,dockerhub,fast api,streamlit,kubernetes,github-actions,prometheus,Grafana,kubeflow
	- AWS Sagemaker.
			
			- sentiment analyzer 
			

			
#####################
Day-10: 4th Jan. 2026
#####################			
			
	MLOps on Cloud ::
	
		- AWS / Azuure 
		
			- AWS Sagemaker!
			
			
			- AWS Cli for Sagemaker Project ::::
			


Project: 

Snapshot of the Source Code -> train.py 
API Web Page 
UI Web Page 
DVC Pipeline 
Prometheus & Grafana 
github-action workflow 
	- Setup the Self-Hosted Runner(K3s) to deploy the model 
	- Setup the Secrets 



PROJECT STRUCTURE (Best Practice)

mlops-e2e/
├── data/
│   ├── raw/
│   │   └── reviews.csv        # NOT tracked by Git
│   └── processed/
│       └── clean.csv          # DVC output
│
├── models/
│   └── model.pkl              # DVC output
│
├── src/
│   ├── data_validation.py
│   ├── preprocessing.py
│   └── train.py
│
├── api/
│   └── main.py
│
├── ui/
│   └── app.py
│
├── dvc.yaml
├── requirements.txt
└── .gitignore
	


🧱 1️⃣ High-Level End-to-End MLOps Architecture
┌────────────┐
│   Developer│
│ (Local PC) │
└─────┬──────┘
      │ git push
      ▼
┌────────────────────┐
│     GitHub Repo    │
│  (Code + DVC meta) │
└─────┬──────────────┘
      │
      │ GitHub Actions (CI/CD)
      ▼
┌──────────────────────────────┐
│        CI PIPELINE           │
│  - Tests                     │
│  - Train Model               │
│  - Log to MLflow             │
│  - Build Docker Images       │
│  - Push to DockerHub         │
└─────┬────────────────────────┘
      │
      ▼
┌──────────────────────┐
│   Kubernetes Cluster │
│   (EC2 / kubeadm)    │
└──────────────────────┘

🧪 2️⃣ Training + Data Versioning Architecture (DVC + MLflow)
           ┌───────────────┐
           │  Raw Dataset  │
           │ (CSV)         │
           └──────┬────────┘
                  │ dvc add
                  ▼
           ┌────────────────┐
           │   DVC Metadata │
           │  (.dvc files)  │
           └──────┬─────────┘
                  │ dvc push
                  ▼
        ┌──────────────────────┐
        │        AWS S3        │
        │ (Raw / Processed /   │
        │  Model Artifacts)    │
        └──────────────────────┘

                  ▲
                  │
┌─────────────────┴─────────────────┐
│           Training Code           │
│  - preprocessing.py               │
│  - train.py                       │
│                                   │
│  Logs → MLflow Tracking Server    │
└─────────────────┬─────────────────┘
                  │
                  ▼
        ┌──────────────────────┐
        │   MLflow Registry    │
        │  sentiment-model     │
        │  v1 → v2 → v3        │
        │  Staging / Prod      │
        └──────────────────────┘


✅ Key idea

DVC → data & model storage

MLflow → model lifecycle

🚀 3️⃣ CI/CD Architecture (GitHub Actions → Docker → Kubernetes)
┌──────────────────────────┐
│      GitHub Actions      │
│                          │
│  Job 1: CI               │
│   - Lint                 │
│   - Unit tests           │
│                          │
│  Job 2: Build            │
│   - Docker API image     │
│   - Docker UI image      │
│                          │
│  Job 3: Push             │
│   - DockerHub            │
│                          │
│  Job 4: Deploy           │
│   - kubectl apply        │
└─────────────┬────────────┘
              │
              ▼
     ┌───────────────────┐
     │    DockerHub      │
     │ api:latest        │
     │ ui:latest         │
     └─────────┬─────────┘
               │
               ▼
     ┌────────────────────────┐
     │   Kubernetes Cluster   │
     │                        │
     │  API Deployment        │
     │  UI Deployment         │
     └────────────────────────┘

☸️ 4️⃣ Kubernetes Runtime Architecture (Pods, Services, Networking)
                   ┌───────────────┐
                   │   User / Web  │
                   └──────┬────────┘
                          │
                     NodePort / LB
                          │
         ┌────────────────┴────────────────┐
         │           Kubernetes            │
         │                                 │
         │  ┌────────────┐   ┌───────────┐ │
         │  │ Streamlit  │   │ FastAPI   │ │
         │  │ UI Pod     │──▶│ API Pod   │ │
         │  └────────────┘   └─────┬─────┘ │
         │                           │
         │                    Load Model
         │                           │
         │                    ┌──────▼──────┐
         │                    │  MLflow     │
         │                    │  Registry   │
         │                    └──────┬──────┘
         │                           │
         │                    ┌──────▼──────┐
         │                    │     S3      │
         │                    │ Model/Data  │
         │                    └─────────────┘
         └──────────────────────────────────┘



















📊 5️⃣ Monitoring Architecture (Prometheus + Grafana)
┌──────────────┐
│   FastAPI    │
│ /metrics     │
└──────┬───────┘
       │ scrape
       ▼
┌──────────────────┐
│   Prometheus     │
│  - Latency       │
│  - Error rate    │
│  - Throughput    │
└──────┬───────────┘
       │
       ▼
┌──────────────────┐
│    Grafana       │
│  Dashboards      │
│  Alerts          │
└──────────────────┘


Git → GitHub Actions → DockerHub → Kubernetes
            ↓
         Training
            ↓
        MLflow Registry
            ↓
      Production Model
            ↓
        FastAPI API
            ↓
        Streamlit UI
            ↓
     Prometheus → Grafana
	 
				
			
🔹 STEP 1: Create AWS Account & Setup
1. Create AWS Account

Sign up at AWS

Enable free tier

2. Create IAM User (IMPORTANT)

Do not use root account.

IAM → Users → Create user

Access type: Programmatic + Console

Attach policies:

AmazonSageMakerFullAccess

AmazonS3FullAccess

CloudWatchFullAccess

Save:

Access Key

Secret Key

🔹 STEP 2: Open SageMaker Studio (First Time)

Go to AWS Console → SageMaker

Click Studio → Set up SageMaker Studio

Choose:

Domain name: sagemaker-demo

Auth: IAM

Create User Profile

Launch SageMaker Studio

⏱️ Takes ~5–10 minutes first time

🔹 STEP 3: Understand SageMaker Components (Very Important)
Component	Purpose
Studio	IDE (Jupyter, terminals)
Notebook	Write & train ML models
S3	Store datasets & models
Training Job	Runs training
Endpoint	Deploy model for inference

👉 SageMaker = Managed ML lifecycle

🔹 STEP 4: Create Your First Notebook

Open SageMaker Studio

Click Notebooks → Python 3 (Data Science)

Create notebook:
01_basic_sagemaker.ipynb

🔹 STEP 5: Upload Dataset to S3
Example: Iris Dataset
import pandas as pd

url = "https://raw.githubusercontent.com/mwaskom/seaborn-data/master/iris.csv"
df = pd.read_csv(url)

df.head()

Upload to S3
import sagemaker
import boto3

session = sagemaker.Session()
bucket = session.default_bucket()

df.to_csv("iris.csv", index=False)
session.upload_data("iris.csv", bucket=bucket, key_prefix="data")

print("Uploaded to:", f"s3://{bucket}/data/iris.csv")

🔹 STEP 6: Train Model Using Built-in SageMaker Algorithm
Use XGBoost (Beginner Friendly)
from sagemaker.inputs import TrainingInput
from sagemaker.xgboost import XGBoost

role = sagemaker.get_execution_role()

xgb = XGBoost(
    entry_point=None,
    framework_version="1.5-1",
    role=role,
    instance_count=1,
    instance_type="ml.m5.large",
    output_path=f"s3://{bucket}/output",
    hyperparameters={
        "objective": "multi:softmax",
        "num_class": 3,
        "max_depth": 5,
        "eta": 0.2,
        "num_round": 100
    }
)

train_input = TrainingInput(
    s3_data=f"s3://{bucket}/data/iris.csv",
    content_type="csv"
)

xgb.fit({"train": train_input})


✔ SageMaker creates:

Training container

Logs in CloudWatch

Model artifact in S3

🔹 STEP 7: Deploy Model (Real Inference)
predictor = xgb.deploy(
    initial_instance_count=1,
    instance_type="ml.m5.large"
)

Test Prediction
import numpy as np

sample = np.array([[5.1, 3.5, 1.4, 0.2]])
predictor.predict(sample)


🎉 You just deployed a real ML API

🔹 STEP 8: Clean Up (Very Important 💸)

Always delete resources!

predictor.delete_endpoint()


Also:

Stop notebook instance

Delete unused S3 files

🔹 STEP 9: Train Using Your Own Script (Intermediate)

Create file train.py

import pandas as pd
from sklearn.ensemble import RandomForestClassifier
import joblib

df = pd.read_csv("/opt/ml/input/data/train/iris.csv")

X = df.iloc[:, :-1]
y = df.iloc[:, -1]

model = RandomForestClassifier()
model.fit(X, y)

joblib.dump(model, "/opt/ml/model/model.joblib")


Then train using Script Mode
(this is how real projects work).

🔹 STEP 10: Real-World SageMaker Flow (Must Know)
Local Dev
   ↓
S3 (data)
   ↓
Training Job
   ↓
Model Artifacts (S3)
   ↓
Endpoint
			
			
			
🔹 What Kubeflow Is (and Why It Fits)

Kubeflow is an open-source MLOps platform built on Kubernetes that supports:

Defining and orchestrating ML pipelines (data preprocess → train → evaluate → deploy).
Kubeflow

Running containerized jobs at each step via Kubernetes.
Medium

Auto-scaling, experiment tracking, artifact logging, model serving (via KServe), hyperparameter tuning (via Katib), etc.
Kubeflow

Because your repository already has:
✔ Kubernetes manifests under k8s/
✔ Docker setup (docker/)
✔ ML code and scripts (src/, models/)
✔ DVC/MLflow tracking — common for MLOps workflows

…these can be adapted to Kubeflow Pipelines by turning various project steps into Kubeflow pipeline components.

🔹 How You Would Deploy This with Kubeflow
✅ 1. Containerize Each Step

Kubeflow Pipelines runs each pipeline step as a container on Kubernetes.
Your existing Dockerfiles can be used for:

Data preprocessing

Model training

Evaluation

Deployment packaging

So you’d wrap each stage as a Kubeflow component (simple Python + KFP SDK).
Medium

✅ 2. Define a Kubeflow Pipeline

Use the Kubeflow Pipelines SDK to write a Python pipeline that:

@dsl.pipeline(...)
def mlops_pipeline(...):
    preprocess = preprocess_op(...)
    train = train_op(...).after(preprocess)
    eval = eval_op(...).after(train)


This pipeline orchestrates your project’s steps in sequence, just like existing scripts or CI/CD workflows do.
Medium

✅ 3. Run on Kubernetes With Kubeflow Installed

Once Kubeflow is installed on your Kubernetes cluster (e.g., Minikube, EKS, GKE, AKS, on-prem etc.), you can:

Upload and run your pipeline via the Kubeflow UI

Watch pipeline runs and artifacts

Scale resources automatically

Kubeflow Pipelines will handle scheduling on Kubernetes.
Kubeflow

✅ 4. Optional Enhancements

You can extend with:

KServe (for scalable model serving)

Katib (for hyperparameter tuning)

Notebooks (for model exploration inside the cluster)

All are supported Kubeflow components.
Kubeflow

🔹 What Kubeflow Does Not Do Automatically

Kubeflow is powerful, but it doesn’t magically automate your entire project without rewriting workflow logic:

❌ It won’t read your current scripts and auto-convert them. You need to define the pipeline components.
❌ It won’t replace MLflow experiment tracking unless you integrate MLflow in your components.
❌ It won’t eliminate the need for Kubernetes expertise. Kubeflow runs on Kubernetes.

🧠 Recommendation Summary
Goal	Possible with Kubeflow?
Orchestrate ML steps on Kubernetes	✅
Track experiments & artifacts	Yes, but you may use Kubeflow metadata + MLflow (hybrid)
Deploy models to production	Yes, via KServe + Kubeflow Pipeline steps
Use current Docker and k8s configs	Reuse them as part of Kubeflow components
📌 Bottom Line

✅ Yes — you can deploy that project using Kubeflow’s MLOps platform.
💡 You’ll convert your existing workflow into Kubeflow pipeline components and run them on Kubernetes with Kubeflow installed.
🛠 You’ll reuse your Docker images and scripts but wrap them with the Kubeflow Pipeline SDK for orchestration.
Kubeflow


🟢 STEP 1: Install k3s

Run on Ubuntu (local VM or EC2)

curl -sfL https://get.k3s.io | sh -


Check:

kubectl get nodes


Expected:

STATUS   Ready

🟢 STEP 2: Install Kubeflow Pipelines (Standalone)

We install ONLY Kubeflow Pipelines, not full Kubeflow (simpler & lighter).

export PIPELINE_VERSION=2.0.5

kubectl apply -k "github.com/kubeflow/pipelines/manifests/kustomize/env/platform-agnostic?ref=${PIPELINE_VERSION}"


Wait:

kubectl get pods -n kubeflow


All pods should be Running.

🟢 STEP 3: Access Kubeflow UI
kubectl port-forward -n kubeflow svc/ml-pipeline-ui 8080:80


Open browser:

http://localhost:8080


You should see Kubeflow Pipelines UI 🎉