MLOps Fundamentals

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Agenda
In this session, we will discuss:
● Introduction to MLOps
● The MLOps lifecycle
● MLOps tools and frameworks
● The MLOps method
● Implemention of MLOps

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps
What is MLOps?
MLOps is the abbreviation for Machine Learning Operations, and it describes the set of practices to
design, deploy, and maintain machine learning in production continuously, reliably, and efficiently.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps

Machine
Learning
MLOps

Data
DevOps
Engineering

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps
Why MLOps?

● Before we develop a machine learning model, we need to think about the business requirements and
added value of our model.
● Combining the data and requirements, we need to find the appropriate algorithm, the recipe.
● During development, we also need a computer, which has its own challenges as well, just like the
kitchen equipment.
● MLOps aims to structure the process and mitigate the risks that are involved in machine learning
development, deployment, and maintenance.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps
The origin of MLOps

MLOps originates from Development Operations, also called DevOps in short. DevOps describes a set of
practices and tools that can be applied to software development to ensure that software is developed
continuously, reliably, and efficiently.

Traditional software development used to be slow because of the separation of Development and
Operations teams.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps
Benefits of MLOps

● Improve the overall speed of developing and delivering machine learning models.
● Processes also become more reliable and secure because of MLOps.
● It aims to bridge the gap between machine learning and operations teams, which enhances
collaboration.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
The MLOps Lifecycle

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
MLOps Lifecycle
● The machine learning lifecycle is one of the fundamental concepts in MLOps.
● It consists of three broad phases:

○ Design
Development Model
○ Design
Development
○ Deployment

Operations

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
MLOps Lifecycle
● Why the machine learning lifecycle?
● Design phase-

○ In the design phase, we focus on the design of the machine learning project.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
MLOps Lifecycle
● Development phase

○ In the development phase, we focus on developing the machine learning model.

● Deployment phase

○ In the deployment phase, we integrate the machine learning model we developed earlier into the
business process.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
MLOps Tools and Frameworks

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Introduction to MLOps Tools and Frameworks
● MLOps tools and frameworks are used to automate the tasks involved in the MLOps lifecycle. This
includes-
○MLFlow
○Kubeflow Pipelines
● There are a number of other MLOps tools available. Some of these tools are more specialized, while
others are more general-purpose. The best tool for you will depend on your specific needs and
requirements.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
 MLOps Method

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
MLOps Method
● The MLOps method is a process that combines machine learning (ML) and
DevOps to automate the end-to-end machine learning lifecycle, from data
preparation to model deployment and monitoring.
● The MLOps method is a continuous process that involves the following
stages:

○ Data preparation

○ Modeling

○ Deployment

○ Monitoring

○ Continuous improvement

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Key principles of MLOps:
● Continuous integration and continuous delivery (CI/CD)
● DevOps culture
● Data as a product
● Monitoring and evaluation

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
 Implementing MLOps

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Implementing MLOps
● The key steps involved in implementing MLOps:

○ Define the business problem

○ Collect and prepare the data

○ Choose the right ML model

○ Train the ML model

○ Deploy the ML model

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Implementing MLOps
● The key steps involved in implementing MLOps:

○ Monitor the ML model

○ Retrain the ML model

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
 Conclusion

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Summary
A brief recap:
● MLOps is a discipline that combines machine learning (ML) and DevOps to automate the end-to-end machine
learning lifecycle, from data preparation to model deployment and monitoring.
● The MLOps lifecycle is a continuous process that includes data preparation, model training, model deployment, and
model monitoring.
● MLOps tools and frameworks are used to automate the tasks involved in the MLOps lifecycle. Some popular MLOps
tools and frameworks include MLFlow, Kubeflow Pipelines.
● The MLOps method is a set of principles and practices that are used to implement MLOps. These principles and
practices include continuous integration and continuous delivery (CI/CD), data as a product, and monitoring and
evaluation.
● Implementing MLOps is the process of putting the MLOps method into practice. This involves: data, modeling,
deployment, and monitoring.

Proprietary content. ©Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.